Agent skill

Curate Project Memory

by hashgraph-online in hashgraph-online/awesome-codex-plugins

Give Codex lasting project context in a local Obsidian-compatible Markdown vault.

Apache-2.0Auto-check passedAgent Workflows

Install Curate Project Memory

skills CLI
$ npx skills add hashgraph-online/awesome-codex-plugins --skill curate-project-memory -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install hashgraph-online/awesome-codex-plugins curate-project-memory --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/hashgraph-online/awesome-codex-plugins.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/DuncanMain/project-memory-core/skills/curate-project-memory .claude/skills/curate-project-memory && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
curate-project-memory
GitHub stars
1.2k
Token cost
~4.7k tokens
SKILL.md length
2,449 words
Files
49 (incl. scripts, references, assets)
Skills in repo
686
Repo updated
First seen
Licence
Apache-2.0

At a glance

Give Codex lasting project context in a local Obsidian-compatible Markdown vault.

  • Works in 7 steps: Explain in one sentence: "PMC gives… → Ask whether to create a new vault or… → After that choice, ask for or confirm… → …
  • A user asks PMC to set up
  • SKILL.md covers First-run experience, Boundaries, Set up a vault and Register a project, plus 18 more sections
  • Connect the projects Obsidian vault

What it does

Curate Project Memory is an agent skill from hashgraph-online/awesome-codex-plugins. Give Codex lasting project context in a local Obsidian-compatible Markdown vault. Use when a user asks PMC to set up or connect the project's Obsidian vault, scan or explain a codebase, create a Project Home, remember a decision, constraint, discovery, procedure, debugging lesson, or current state, recall what Codex already knows, prepare a handoff, check whether notes are stale or contradictory, relate code to knowledge, keep something session-only, export or share reviewed context, or resume work from existing…

Its SKILL.md is about 4.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 51 other files, including scripts, reference files and assets (for example `agents/openai.yaml`, `assets/agents-project-memory.md` and `assets/constraint.md`).

It sits in Agent Workflows, covering Agent memory and Session handoff. It works with Obsidian. The repository describes itself as: A curated list of awesome OpenAI Codex / ChatGPT plugins, skills, and resources. The 1 Codex Marketplace. See live plugins at: https://hol.org/plugins/best-codex-plugins. The licence is Apache-2.0.

When your agent uses it

  • A user asks PMC to set up
  • Connect the projects Obsidian vault
  • Explain a codebase
  • Create a Project Home

Example prompts

  • “/curate-project-memory”

Workflow steps

7 steps, taken from the first numbered list in SKILL.md.

  1. Explain in one sentence: "PMC gives Codex lasting project context in an Obsidian vault you control."
  2. Ask whether to create a new vault or connect an existing Obsidian vault. Ask only this one question first.
  3. After that choice, ask for or confirm the vault path, infer the current repository, and propose a readable project name.
  4. Complete the vault setup and project registration conversationally. Do not require the user to know commands, markers, folder layouts, or…
  5. Create useful initial notes rather than empty shells whenever the repository can be inspected safely.
  6. At completion, state where the vault and project notes live, explain how to open the vault in Obsidian, and offer to scan the repository…
  7. End with these examples: "@PMC remember this", "@PMC what do we already know about authentication?", and "@PMC wrap up this task".

What it can do on your machine

Read from SKILL.md and the folder at commit 78497e5. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/, which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Curate Project Memory loads about 4.7k tokens when it runs, and up to ~12k if it reads all its reference files. Until then it costs about 139 tokens; SKILL.md has 2,449 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~139
When it runs · the whole SKILL.md, loaded when a task matches
~4.7k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~12k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from hashgraph-online/awesome-codex-plugins at commit 78497e5, republished under its Apache-2.0 licence (© hashgraph-online). 2,449 words, ~4,678 tokens.

Download SKILL.mdSave it as .claude/skills/curate-project-memory/SKILL.md (or your agent's skills folder). This skill also uses 48 other files; get the full folder from GitHub.
name
curate-project-memory
description
Give Codex lasting project context in a local Obsidian-compatible Markdown vault. Use when a user asks PMC to set up or connect the project's Obsidian vault, scan or explain a codebase, create a Project Home, remember a decision, constraint, discovery, procedure, debugging lesson, or current state, recall what Codex already knows, prepare a handoff, check whether notes are stale or contradictory, relate code to knowledge, keep something session-only, export or share reviewed context, or resume work from existing project notes.

Curate Project Memory

Use conversation as the interface. Do not ask the user to run commands or edit configuration manually. Use Codex's normal file tools and request narrowly scoped filesystem permission when needed.

First-run experience

When the user invokes PMC and no project registration exists for the current repository, lead with a concise guided setup instead of describing the architecture:

  1. Explain in one sentence: "PMC gives Codex lasting project context in an Obsidian vault you control."
  2. Ask whether to create a new vault or connect an existing Obsidian vault. Ask only this one question first.
  3. After that choice, ask for or confirm the vault path, infer the current repository, and propose a readable project name.
  4. Complete the vault setup and project registration conversationally. Do not require the user to know commands, markers, folder layouts, or configuration formats.
  5. Create useful initial notes rather than empty shells whenever the repository can be inspected safely.
  6. At completion, state where the vault and project notes live, explain how to open the vault in Obsidian, and offer to scan the repository now.
  7. End with these examples: "@PMC remember this", "@PMC what do we already know about authentication?", and "@PMC wrap up this task".

Do not overwhelm a first-time user with advanced features. Introduce context packs, health checks, exports, team review, and the companion only when relevant.

Read references/project-profiles.md when choosing registration defaults.

Read references/local-similarity.md before building or querying the optional machine-local similarity index.

Read references/storage-and-portability.md for setup or project registration. Read references/automatic-orientation.md when the user wants the vault added as a project source or wants future chats to load project notes without a prompt. Read references/durable-note-policy.md before promoting, changing, or removing knowledge. Read references/note-structure.md when creating notes. Read references/context-packs.md when orienting or resuming work. Read references/knowledge-health.md before auditing a vault. Read references/code-traceability.md when relating notes to source files. Read references/branch-awareness.md before promoting knowledge discovered on a branch or worktree. Read references/drift-detection.md before creating or running verification checks. Read references/coverage.md before assessing documentation coverage. Read references/team-collaboration.md before configuring shared-vault ownership or reviews. Read references/exports.md before creating a portable brief or audit artifact. Read references/obsidian-companion.md when installing, configuring, or responding to review actions from the optional Obsidian companion.

Boundaries

  • Operate only on local paths that the user selects or places in scope.
  • Do not upload, transmit, or add telemetry.
  • Do not request or capture complete raw chat transcripts.
  • Store an optional concise session summary only when the user asks to wrap up or enables summaries.
  • Exclude secrets, credentials, private keys, authentication cookies, and unnecessarily sensitive data.
  • Treat the vault as curated knowledge, not a chat archive.
  • Never contradict, delete, or supersede durable knowledge silently.
  • Do not require Obsidian community plugins.
  • If local filesystem access is unavailable, explain that setup requires a local Codex environment; do not pretend files were created.

Set up a vault

Determine whether the user clearly requested a new vault or an existing vault. If ambiguous, ask one short question.

New vault
  1. Confirm the path and request permission if necessary.
  2. Refuse to repurpose a populated non-Obsidian folder without explicit confirmation.
  3. Create the selected folder and an empty .obsidian directory.
  4. Create only the minimal shared structure from references/note-structure.md.
  5. Create the machine-local configuration described in references/storage-and-portability.md.
  6. Report the created paths and explain that the folder can now be opened in Obsidian.
Existing vault
  1. Confirm that the folder exists.
  2. Treat .obsidian as evidence that it is an Obsidian vault.
  3. If .obsidian is absent, explain that it appears to be a Markdown folder and ask before using it.
  4. Preserve the existing organisation. Do not move or rename existing content unless asked.
  5. Add project-memory notes incrementally and only where useful.

Register a project

  1. Infer the repository root from the current workspace when possible.
  2. Ask for a project name and vault folder only when they cannot be inferred safely.
  3. Create a stable lower-case project ID that does not depend on drive letters.
  4. Store the absolute repository path only in machine-local configuration.
  5. Store repository-relative code references in durable notes.
  6. Create Project Home.md, Project.md, and Current State.md from the bundled templates when missing. Project Home.md is the attractive human-facing front door; Project.md remains the compact compatibility index used by existing workflows. Do not create empty directory forests.
  7. Create Inbox/Promotion Inbox.md, Handoff.md, Coverage.md, and the useful Bases dashboards described in references/note-structure.md unless the user opts out.
  8. Offer automatic project orientation as an opt-in. If accepted, follow references/automatic-orientation.md and preserve existing repository instructions.
  9. Infer or ask for an optional project profile only when it improves initial coverage areas. Do not create empty folders from a profile.

Create the initial project orientation

Trigger on requests such as "create an overview of this codebase", "scan this project", or acceptance of the setup offer.

  1. Inspect the repository's primary README, package or build metadata, repository instructions, top-level structure, Git status, and a small number of entry points relevant to understanding the project.
  2. Summarize what the project does, how it is organised, how to run or test it, its current state, important constraints, and likely next actions. Distinguish verified facts from inference.
  3. Populate Project Home.md, Project.md, and Current State.md; create focused decision, constraint, discovery, procedure, or debugging-lesson notes only when evidence supports them.
  4. Link useful notes from Project Home.md and keep it concise enough to scan quickly in Obsidian.
  5. Report the evidence inspected and the exact vault files created or updated.

Use the vault as a project source

For a Codex local project, guide the user through the one-time UI action: open Edit project, select Add folder, add the vault, and keep the code repository as the primary folder. Do not claim to have attached a folder through the UI.

Treat the vault as a secondary source folder. Read and edit only relevant notes. Secondary folders do not automatically contribute their own AGENTS.md, skills, or project configuration, so put automatic-orientation guidance in the primary code repository after explicit opt-in.

Enable automatic project orientation

  1. Confirm that the user wants relevant durable notes read at the beginning of future tasks for this repository.
  2. Copy the managed block from assets/agents-project-memory.md into the primary repository's AGENTS.md; create that file only when necessary.
  3. Replace the project ID token and preserve all unrelated instructions.
  4. Do not include an absolute vault or repository path in AGENTS.md.
  5. On future tasks, load Project.md and Current State.md first, then follow only relevant links.
  6. If the vault is unavailable or permission is missing, continue safely and state that project-memory context was not loaded.

Recognize markers

Interpret natural equivalents, not only exact prefixes:

  • Remember: -> confirmed durable candidate.
  • Decision: -> decision candidate with rationale and consequences.
  • Constraint: -> requirement or constraint candidate.
  • Reference: -> durable reference candidate.
  • Session only: -> do not promote to durable notes.
  • Forget: -> propose correction, removal, or supersession and require confirmation.

Treat these natural requests as first-class equivalents:

  • @PMC remember this -> summarize the immediately preceding durable point and ask for confirmation when its scope is ambiguous.
  • @PMC remember the decision we just made -> create or update a decision after conflict and evidence checks.
  • @PMC update the project status -> propose a concise Current State.md update.
  • @PMC what do we already know about X? -> build a focused context pack and answer with note links.
  • @PMC wrap up this task -> follow the wrap-up workflow.

Markers never bypass secret filtering or conflict checks.

Prompt for knowledge worth keeping

Do not rely on the user to remember a special marker. During ordinary project work, notice information that is likely to matter in a future task, especially:

  • an accepted architectural or product decision and its rationale;
  • a durable constraint, invariant, convention, or non-obvious requirement;
  • a verified discovery that changes how the project should be understood;
  • a reusable procedure, workaround, or operational lesson;
  • a change that makes an existing durable note stale or incorrect.

At a natural checkpoint, ask one concise question: "This will matter in future tasks. Add it to the project's Obsidian vault?" Include a one-line summary of exactly what would be recorded. Do not interrupt for routine edits, speculative ideas, transient debugging output, or facts already captured. Batch multiple related candidates into one review prompt when practical.

If the user says yes, treat that response as an explicit durable marker and follow the normal conflict, branch, evidence, and secret checks before writing. If the user says no, keep it session-only and do not repeatedly ask about the same candidate. When wrapping up, always include a short separate list titled Worth adding to the project's Obsidian vault or state that no durable candidates were found.

Show full SKILL.md (1,013 more words)Show less

Promote durable knowledge

  1. Extract only information likely to help future project work.
  2. Read relevant existing notes before proposing a change.
  3. Classify each candidate as create, update, duplicate, conflict, supersede, or session-only.
  4. Add unmarked candidates to Inbox/Promotion Inbox.md using assets/promotion-inbox.md. Record the proposed operation, target, evidence, confidence, and conflict status without treating the candidate as durable truth. For a conflict, quote concise existing and proposed claims with separate evidence so the reviewer can compare both sides.
  5. For an explicit marker, write after checking for conflicts unless the user requested review-only mode.
  6. Ask before resolving a conflict, deletion, or supersession.
  7. When reviewing the inbox, let the user approve, edit, reject, or defer each candidate independently. Follow the guarded lifecycle in references/obsidian-companion.md when applying an approved candidate, record the outcome, and keep rejected candidates out of durable notes.
  8. Report the exact durable files changed.

Before promotion, inspect the repository branch and revision when Git is available. Treat branch-only discoveries according to references/branch-awareness.md; never present unmerged implementation as canonical released state.

Build a context pack

  1. Read references/context-packs.md.
  2. Run scripts/build_context_pack.py <project-folder> --query <task> yourself to produce a deterministic candidate ranking; pass each relevant repository-relative path with --code-path.
  3. Always inspect the project's Project.md and Current State.md when available, then inspect only the highest-value ranked candidates required for safe work.
  4. Exclude rejected, deprecated, and superseded knowledge except when historical context is necessary.
  5. State briefly which additional notes were loaded and why. Never load the entire vault by default.
  6. If terminology mismatch leaves important knowledge undiscovered, offer the optional local similarity index from references/local-similarity.md; keep its generated index outside the vault and inspect every returned note.

Trace code and knowledge

Use the schema in references/code-traceability.md. Store stable repository IDs and repository-relative paths, never drive-specific paths. When code changes implement or invalidate durable knowledge, propose reciprocal links or status updates through the Promotion Inbox. Answer "which decisions affect this file?" by running the context-pack builder with --code-path and verifying the returned notes.

Review staleness and contradictions

Use the health checker for objective structural signals, then compare potentially affected claims with current code, configuration, tests, and accepted notes. Record suspected contradictions in the Promotion Inbox using Existing claim, Existing evidence, Proposed change, and Evidence. Do not choose a winner or rewrite canonical notes silently. Distinguish stale, conflicting, and unverified; a recent timestamp is not proof of correctness.

Detect evidence drift

Read references/drift-detection.md. Add verification checks only for important claims that have stable, non-sensitive machine-verifiable evidence. Run scripts/check_drift.py <project-folder> <repository-root> yourself. Inspect failed checks, distinguish a changed implementation from a bad check, and place proposed corrections in the Promotion Inbox. Never auto-rewrite canonical knowledge from a failed check.

Assess knowledge coverage

Read references/coverage.md. Run scripts/knowledge_coverage.py <project-folder> yourself, passing expected workstreams from Project metadata when needed. Update Coverage.md from assets/coverage.md with the results and links. Treat coverage as a navigation and review signal, not a performance score.

Collaborate in a team vault

Read references/team-collaboration.md. Use one durable decision per file, stable knowledge_id values, explicit owners and reviewers, and contributor-specific inboxes when several people work concurrently. Never infer approval merely from a Git merge. Resolve semantic conflicts through reviewed supersession or correction, not by keeping both claims active.

Export project knowledge

Read references/exports.md. Run the health checker first and stop if possible-secret findings remain. Run scripts/export_project.py <project-folder> <output-file> --mode <onboarding|decision-log|audit|context> yourself. For context mode, pass only paths already selected and inspected for the current task. Do not overwrite an existing export without the user's approval. Explain that exports are snapshots and do not update automatically.

Interoperate with the Obsidian companion

Read references/obsidian-companion.md. Treat companion review actions as workflow state, not proof that a proposed semantic change has been applied. When asked to apply approved candidates, run scripts/candidate_lifecycle.py to verify the approval fingerprint and perform guarded state transitions, rerun current conflict and evidence checks, update canonical notes, and record applied, conflict, or failed with outcome metadata. Never assume approval resolved a conflict automatically.

Check knowledge health

Run scripts/check_vault_health.py <project-folder> yourself when the user asks to audit, review, or check the vault. Do not ask the user to use a terminal. Interpret its JSON report using references/knowledge-health.md; inspect reported notes before proposing repairs. When the user wants portable or companion-visible results, run scripts/generate_health_report.py <project-folder> <project-folder>/Health Report.md --project-id <id> and include --repository-root when available. Never make destructive repairs automatically.

For "@PMC check this project's memory" or "check whether these notes are out of date", combine structural health with a focused evidence review:

  1. Run the health checker.
  2. Inspect Project Home.md, Project.md, Current State.md, active decisions, and unresolved Promotion Inbox items.
  3. Compare claimed commands, paths, status, dependencies, and completed work with current repository evidence where safe.
  4. Report stale information, contradictions, missing setup knowledge, broken links, completed actions still shown as active, and important repository changes not reflected in the vault.
  5. Separate safe mechanical repairs from semantic changes. Ask before changing meaning, resolving conflicts, deleting, or superseding knowledge.

Create a handoff brief

On wrap-up, project restart, or explicit handoff request, update Handoff.md from assets/handoff.md. Keep it one-page in spirit: objective, current position, recent work, accepted decisions, blockers, open questions, next actions, and relevant links. Treat it as a generated navigation aid, not a replacement for canonical notes.

Wrap up

Trigger on requests such as "wrap up", "finish for today", or "summarize this session".

  1. Summarize the objective, work completed, decisions, discoveries, changed files, open questions, and next actions.
  2. Omit transient tool output, secrets, and unnecessary conversation detail.
  3. If session summaries are enabled or requested, save the concise summary outside the Obsidian vault using the configured session-summary location.
  4. Propose durable updates separately from the summary.
  5. Apply only approved durable changes.
  6. Update Handoff.md after approved durable changes when it would materially help the next session.
  7. Never imply that a complete transcript was archived.

Resume work

Read Project Home.md, Project.md, Current State.md, and only the relevant linked notes. Summarize current status, known constraints, unresolved questions, and likely next steps. Flag stale or contradictory notes instead of guessing which is correct.

© hashgraph-online, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 48 other files (scripts, references, assets) in plugins/DuncanMain/project-memory-core/skills/curate-project-memory of hashgraph-online/awesome-codex-plugins.

  • SKILL.md
  • agents/openai.yaml
  • assets/agents-project-memory.md
  • assets/bases/Active Decisions.base
  • assets/bases/Code Traceability.base
  • assets/bases/Current Work.base
  • assets/bases/Knowledge Health.base
  • assets/bases/Team Review Queue.base
  • assets/bases/Verification and Coverage.base
  • assets/constraint.md
  • assets/coverage.md
  • assets/current-state.md
  • assets/debugging-lesson.md
  • assets/decision.md
  • assets/discovery.md
  • assets/handoff.md
  • assets/procedure.md
  • assets/project-home.md
  • … and 31 more

Open the folder on GitHubat commit 78497e5

Compare with similar skills

Curate Project Memory next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Curate Project Memory compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Curate Project Memory this skillhashgraph-online/awesome-codex-plugins1.2k—~4.7kAutomated safety check: PassApache-2.0
Coding Agent Session Findercode-yeongyu/oh-my-openagent70k1 repos~2.8kAutomated safety check: PassCustom licence
Beads Task Memorygastownhall/beads28k—~1.2kAutomated safety check: PassMIT
Session History Searchslopus/happy24k—~3.1kAutomated safety check: PassMIT
Memori Long-Term MemoryMemoriLabs/Memori17k—~2kAutomated safety check: NotesCustom licence
Mindmemos CLImindscale-noah/MindMemOS1k1 repos~3.4kAutomated safety check: PassNone

Similar skills

  • Coding Agent Session Finder

    code-yeongyu/oh-my-openagent

    Finds, reads and reconstructs past coding-agent sessions across Codex, Claude, OpenCode, Senpi and many other local agent logs.

    70k GitHub starsUsed in 1 repo~2.8k tokens
    Agent WorkflowsAuto-check passed
  • Beads Task Memory

    gastownhall/beads

    Tracks multi-session work with dependencies in the bd issue tracker so the agent can find ready tasks and recover its context after conversation compaction.

    28k GitHub stars~1.2k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Searches past Claude Code, Codex and Cursor sessions and summarizes what was worked on, tried or decided, using extraction scripts instead of reading raw logs.

    24k GitHub stars~3.1k tokensUpdated yesterday
    Agent WorkflowsAuto-check passed
  • Memori Long-Term Memory

    MemoriLabs/Memori

    Connects Claude Code to Memori Cloud for long-term memory, recalling stored context before substantive replies and saving new context afterward.

    17k GitHub stars~2k tokensUpdated 5 days ago
    Agent WorkflowsAuto-check: notes
  • Mindmemos CLI

    mindscale-noah/MindMemOS

    Give an AI agent persistent, cross-session long-term memory through MindMemOS.

    1k GitHub starsUsed in 1 repo~3.4k tokens
    Agent WorkflowsAuto-check passed
  • Qmd

    breferrari/obsidian-mind

    Search the vault using QMD semantic search. An agent skill from breferrari/obsidian-mind.

    4.9k GitHub stars~1.7k tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed

More from hashgraph-online/awesome-codex-plugins

All 686 skills in this repo
  • Anime Reaction Gif

    hashgraph-online/awesome-codex-plugins

    Create original anime-style reaction stickers as looping GIFs and MP4 previews, using generated character pose sheets and timed key poses.

    1.2k GitHub stars~922 tokensUpdated today
    Auto-check passed
  • Calibredb

    hashgraph-online/awesome-codex-plugins

    Manage and query Calibre libraries with the calibredb CLI (local paths or Calibre Content server URLs).

    1.2k GitHub stars~1k tokensUpdated today
    Auto-check passed
  • Rust API Test Harness

    hashgraph-online/awesome-codex-plugins

    A skill your agent uses when adding, changing, testing, or debugging Rust HTTP APIs and services, especially when Codex needs black-box integration tests, random-port app startup, real database test…

    1.2k GitHub stars~1.7k tokensUpdated today
    Auto-check passed
  • Art

    hashgraph-online/awesome-codex-plugins

    Make a studio's game look like something at build time — a cover from a real frame of the game (free), painted covers, backdrops, textures and character plates from image models through the…

    1.2k GitHub stars~2.4k tokensUpdated today
    Auto-check passed
  • Game Balance Economy

    hashgraph-online/awesome-codex-plugins

    Balance game difficulty, resources, rewards, probability, progression, economies, and dominant strategies.

    1.2k GitHub stars~618 tokensUpdated today
    Auto-check passed
  • Manuscript Engagement Analytics

    hashgraph-online/awesome-codex-plugins

    Analyze nonfiction manuscripts for reader engagement signals, including heading-level word counts, slow starts, long slogs, weak takeaway titles, value pacing, beta-reader comment dropoff, and…

    1.2k GitHub stars~875 tokensUpdated today
    Auto-check passed

Works with

Categories

Questions about Curate Project Memory

What does Curate Project Memory do?

Give Codex lasting project context in a local Obsidian-compatible Markdown vault. Curate Project Memory is an agent skill from hashgraph-online/awesome-codex-plugins. Give Codex lasting project context in a local Obsidian-compatible Markdown vault.

When should I use Curate Project Memory?

Curate Project Memory fits situations like: A user asks PMC to set up; connect the projects Obsidian vault; explain a codebase; create a Project Home.

How do I install Curate Project Memory in Claude Code?

Run `npx skills add hashgraph-online/awesome-codex-plugins --skill curate-project-memory -a claude-code`. Or copy the skill folder (plugins/DuncanMain/project-memory-core/skills/curate-project-memory in hashgraph-online/awesome-codex-plugins) into .claude/skills/curate-project-memory in your project. Claude Code loads it when a task matches its description.

How do I install Curate Project Memory in Codex?

Run `npx skills add hashgraph-online/awesome-codex-plugins --skill curate-project-memory -a codex`. Or copy the skill folder (plugins/DuncanMain/project-memory-core/skills/curate-project-memory in hashgraph-online/awesome-codex-plugins) into .agents/skills/curate-project-memory in your project. Codex loads it when a task matches its description.

Can I use Curate Project Memory in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add hashgraph-online/awesome-codex-plugins --skill curate-project-memory -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/curate-project-memory, .gemini/skills/curate-project-memory, .github/skills/curate-project-memory and .opencode/skills/curate-project-memory in your project.

What does Curate Project Memory need to run?

SKILL.md names no scripts, command-line tools or credentials: Curate Project Memory is instructions for the agent only.

Does Curate Project Memory access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Curate Project Memory safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Curate Project Memory use?

Curate Project Memory is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Curate Project Memory use?

About 4.7k tokens (SKILL.md is roughly 19k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 7k tokens, read only when the agent opens those files.

What are the alternatives to Curate Project Memory?

Skills that share tags, products or a category with Curate Project Memory: Coding Agent Session Finder (code-yeongyu/oh-my-openagent, 70k stars), Beads Task Memory (gastownhall/beads, 28k stars), Session History Search (slopus/happy, 24k stars) and Memori Long-Term Memory (MemoriLabs/Memori, 17k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Curate Project Memory?

hashgraph-online (a GitHub organization) maintains it in hashgraph-online/awesome-codex-plugins, which has 1,242 GitHub stars. The repository holds 686 skills in this directory. The repository was last updated on October 8, 2026.

Source: hashgraph-online/awesome-codex-plugins on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.