New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Tune Temperature Policy

by radimsem in radimsem/remindb

A skill your agent uses when changing any field in pkg/temperature/Config (DecayRate, AccessBoost, ColdThreshold, NotifyThreshold, TickInterval) or modifying decayFactor / Score / cold-node…

MITAuto-check passedAgent Workflows

Install Tune Temperature Policy

skills CLI
$ npx skills add radimsem/remindb --skill tune-temperature-policy -a claude-code

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

GitHub CLI
$ gh skill install radimsem/remindb tune-temperature-policy --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/radimsem/remindb.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/tune-temperature-policy .claude/skills/tune-temperature-policy && 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
tune-temperature-policy
GitHub stars
129
Token cost
~2.3k tokens
SKILL.md length
946 words
Files
1
Skills in repo
11
Repo updated
First seen
Licence
MIT

At a glance

A skill your agent uses when changing any field in pkg/temperature/Config (DecayRate, AccessBoost, ColdThreshold, NotifyThreshold, TickInterval) or modifying decayFactor / Score / cold-node…

  • Changing any field in pkg/temperature/Config (DecayRate
  • SKILL.md covers What the knobs do, Where the change ripples, Tuning rationales — what to… and The docs-sync step, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
  • NotifyThreshold

What it does

Tune Temperature Policy is an agent skill from radimsem/remindb. Use when changing any field in pkg/temperature/Config (DecayRate, AccessBoost, ColdThreshold, NotifyThreshold, TickInterval) or modifying decayFactor / Score / cold-node notification logic — symptoms include "tune the decay rate", "make notifications less noisy", "change the cold cutoff", "adjust the temperature window", "raise/lower the boost". Prevents silent docs drift in skills/remind/SKILL.md (mental-model numerics) and skills/memorize/references/lifecycle.md (summarization workflow triggered by the…

Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Summarization. It works with Model Context Protocol. The repository describes itself as: An agentic memory database that cuts session tokens by 82–99%. One portable SQLite file — your agent's memory, anywhere. The licence is MIT.

When your agent uses it

  • Changing any field in pkg/temperature/Config (DecayRate
  • NotifyThreshold
  • Modifying decayFactor / Score / cold-node notification logic — symptoms include tune the decay rate
  • Make notifications less noisy

Example prompts

  • “tune the decay rate”
  • “make notifications less noisy”
  • “change the cold cutoff”
  • “/tune-temperature-policy”

What it can do on your machine

Read from SKILL.md and the folder at commit 977b31c. 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

    No scripts in the folder and no shell commands in SKILL.md.

    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

Tune Temperature Policy loads about 2.3k tokens when it runs. Until then it costs about 142 tokens; SKILL.md has 946 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~142
When it runs · the whole SKILL.md, loaded when a task matches
~2.3k

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); files beside SKILL.md are not scanned.

SKILL.md

The full file from radimsem/remindb at commit 977b31c, republished under its MIT licence (© radimsem). 946 words, ~2,293 tokens.

Download SKILL.mdSave it as .claude/skills/tune-temperature-policy/SKILL.md (or your agent's skills folder).
name
tune-temperature-policy
description
Use when changing any field in `pkg/temperature/Config` (`DecayRate`, `AccessBoost`, `ColdThreshold`, `NotifyThreshold`, `TickInterval`) or modifying `decayFactor` / `Score` / cold-node notification logic — symptoms include "tune the decay rate", "make notifications less noisy", "change the cold cutoff", "adjust the temperature window", "raise/lower the boost". Prevents silent docs drift in `skills/remind/SKILL.md` (mental-model numerics) and `skills/memorize/references/lifecycle.md` (summarization workflow triggered by the notification).

Tune the temperature policy

remindb's temperature system has five knobs in pkg/temperature/config.go. They're tightly coupled — changing one shifts the behavior of search ranking, the cold-set query, and the client-facing notification stream. Tuning is rarely a one-file change.

The skill exists because two public skills document this policy for agents:

  • skills/remind/SKILL.md — owns the numerics in the mental model (decay rate, access boost, cold/notify thresholds, tick interval, ranking score formula, notification payload shape). These stay in the SKILL.md router, not a reference.
  • skills/memorize/references/lifecycle.md — owns the summarization workflow the notification triggers (MemoryFetch → MemorySummarize). The memorize SKILL.md router only points at it.

If the numbers or the workflow drift from the code, agents will reason from stale defaults.

What the knobs do

KnobDefaultAffects
DecayRate0.05decayFactor = exp(-rate × elapsed_hours) — applied each tick to every node
AccessBoost0.15Added to a node's temperature on read; capped at 1.0 by SQL min(1.0, …)
ColdThreshold0.1Below this, nodes are "cold" — used by GetColdNodes and the search relevance floor (score = relevance × (0.3 + 0.7 × temperature) × recency)
NotifyThreshold0.1Below this, the server pushes an MCP notification (level: "warning", logger: "remindb.temperature") — gated by per-node hysteresis dedup
TickInterval5 * time.MinuteHow often Tracker.Run decays + queries cold nodes

Where the change ripples

Every tune touches four surfaces minimum.

FileWhy
pkg/temperature/config.goThe knob itself (the DefaultConfig literal)
pkg/temperature/*_test.goTests that assert specific numeric outcomes (tracker_test.go, cold_test.go, decay_test.go) — they'll fail if defaults shift
pkg/mcp/server_test.goIf NotifyThreshold semantics change, the dedup/hysteresis tests need updating
skills/remind/SKILL.mdPublic-facing mental-model docs (numerics, ranking score, notification payload, threshold descriptions) — the easy one to forget
skills/memorize/references/lifecycle.mdPublic-facing summarization workflow triggered by the notification (MemoryFetch → MemorySummarize) — touch when the trigger or the recommended response shape changes

If the change is structural (new knob, new threshold), also add a note to pkg/temperature/cold.go and re-read pkg/mcp/server.go:60-98 (NotifyColdNodes / selectNewNotifications) to confirm the hysteresis logic still makes sense.

Tuning rationales — what to tune for what symptom

SymptomKnob to considerDirection
Cold notifications too noisyNotifyThresholdLower (e.g., 0.05) — only the very coldest get pushed; widens the hysteresis band so re-notifications are rarer
Cold notifications too rareNotifyThresholdRaise toward ColdThreshold
Hot nodes lingering at the top of searchDecayRateRaise (e.g., 0.1) — decay is faster, ranking turnover is quicker
Recent reads not boosting enoughAccessBoostRaise (e.g., 0.25) — fewer reads needed to keep a node warm
Tick storms (decay bursts visible in logs)TickIntervalRaise (e.g., 15 min) — fewer, larger decays per tick (factor stays the same since it's based on elapsed hours)
Cold-set query returning too much / too littleColdThresholdAdjust to match what GetColdNodes should return

Note the asymmetry: ColdThreshold and NotifyThreshold can be different. Currently they're both 0.1 so the cold-set and the notify-set are the same; setting NotifyThreshold < ColdThreshold gives you a "cold but not yet alertable" zone.

The docs-sync step

The two public skills carry different surfaces of the policy. Walk both:

skills/remind/SKILL.md — mental-model numerics
  • Frontmatter description — mentions "warning-level cold-node notifications"
  • Mental model → Nodes — quotes +0.15, exp(-0.05 × elapsed_hours), ~5% per hour, the two thresholds, and 0.1 defaults
  • Mental model → Ranking — score = relevance × (0.3 + 0.7 × temperature) × recency
  • Mental model → Notifications — quotes the message string, hysteresis behavior, payload shape
  • Anti-patterns — the dedup-and-rearm note, the ColdThreshold vs NotifyThreshold distinction
skills/memorize/references/lifecycle.md — workflow that follows the notification
  • Summarize a cold node — the notification handoff — the MemoryFetch → MemorySummarize flow. Touch when the trigger semantics, the recommended summary shape, or MemorySummarize's preserved-fields contract changes. (memorize/SKILL.md only carries the one-line playbook row + the pointer.)
  • Maintenance cadence — the "on a remindb.temperature warning → summarize" entry that frames when to reach for the workflow.
Show full SKILL.md (360 more words)Show less
Walking the change

Every numeric or behavioral change requires a pass through both skills. If you change DecayRate from 0.05 to 0.1, every 0.05 and "5% per hour" must update in remind's SKILL.md. If you decouple ColdThreshold and NotifyThreshold, the threshold paragraphs in remind's SKILL.md need updating. If you change what MemorySummarize preserves, memorize's references/lifecycle.md summarize section needs updating.

The fast check (grep recursively — depth lives in references/):

grep -rnE '0\.05|0\.15|0\.1|5 min' skills/remind/
grep -rnE 'MemorySummarize|NotifyThreshold|ColdThreshold' skills/remind/ skills/memorize/

Every hit is a candidate for an update.

Quick reference

1. pkg/temperature/config.go               (the knob)
2. pkg/temperature/*_test.go               (assertions on numerics)
3. pkg/mcp/server_test.go                  (only if NotifyThreshold semantics change)
4. skills/remind/SKILL.md                  (mental-model numerics + behavioral descriptions)
5. skills/memorize/references/lifecycle.md  (only if the summarization workflow or MemorySummarize contract changes)
6. go test ./pkg/temperature/... ./pkg/mcp/...    (must pass)

Common mistakes

  • Changing the default but not the test that asserts it. tracker_test.go:103 and cold_test.go check specific decay outcomes from the default config. If you bump DecayRate, the expected post-tick temperatures must change too.
  • Expecting NotifyThreshold > ColdThreshold to alert on warmer nodes. It doesn't. The cold set is gated upstream at ColdThreshold in Tracker.Tick; NotifyThreshold only filters within that set via n.Temperature >= s.notifyThreshold in selectNewNotifications. Setting NotifyThreshold above ColdThreshold just disables the filter — every node already in the cold set passes through. To widen the alerting set, raise ColdThreshold. To narrow it, lower NotifyThreshold below ColdThreshold (creates a "cold but not alertable" hysteresis band).
  • Skipping the public-skill docs sync. Drift between the code and either skills/remind/SKILL.md (numerics) or skills/memorize/references/lifecycle.md (summarization workflow) means a future Claude reasons from a stale baseline. Both skills are part of the deployed surface; treat drift as a bug.
  • Leaving boostResultNodes calls in mutating MCP tools. Boost is for read tools (the read is the access). If you raise AccessBoost and a write tool also boosts, mutations look like accesses and skew temperatures up. Audit pkg/mcp/tools/ after raising the boost.
  • Bumping TickInterval without thinking about hysteresis. Notifications dedup per-node-per-cold-state. A longer tick means longer between dedup-eviction opportunities; a node oscillating around NotifyThreshold may go quieter than expected.

Cross-references

  • .claude/rules/go-concise.md — error handling, named locals
  • .claude/skills/add-mcp-tool/SKILL.md — for the boostResultNodes rule when adding new tools (so the boost contract stays clean)
  • skills/remind/SKILL.md — read-side docs target (numerics, ranking score, notification payload, threshold descriptions)
  • skills/memorize/references/lifecycle.md — write-side docs target (the cold-node summarization workflow and MemorySummarize contract)
  • pkg/temperature/decay.go — the Score formula constants (coldFloor = 0.3, tempWeight = 0.7); these are not in Config but they shape ranking and may need to move there if you tune them

© radimsem, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/tune-temperature-policy of radimsem/remindb.

Open the folder on GitHubat commit 977b31c

Compare with similar skills

Tune Temperature Policy 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.

Tune Temperature Policy compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Tune Temperature Policy this skillradimsem/remindb129—~2.3kAutomated safety check: PassMIT
Vss MCP Integrationopen-edge-platform/edge-ai-libraries169—~1.5kAutomated safety check: NotesApache-2.0
Spark Reportbreakstageaxe61/genspark-claw159—~738Automated safety check: PassMIT
Tldr Promptgithub/awesome-copilot40k1 repos~3.1kAutomated safety check: PassMIT
Cloud API Recipe AuthoringTencentCloudBase/CloudBase-AI-Toolkit1.1k—~3.7kAutomated safety check: PassMIT
Gearcoleco Romhackingdrhelius/Gearcoleco142—~3.9kAutomated safety check: PassGPL-3.0

Similar skills

  • Vss MCP Integration

    open-edge-platform/edge-ai-libraries

    Helps developers configure and extend the Video Search and Summarization sample app's spec-driven FastMCP proxy for VSS search.

    169 GitHub stars~1.5k tokensUpdated today
    Agent WorkflowsAuto-check: notes

  • Spark Report

    breakstageaxe61/genspark-claw

    Turn raw research, notes, links, or a conversation into a polished, shareable Sparkpage-style report page with an executive summary, structured sections, tables, and citations.

    159 GitHub stars~738 tokensUpdated 13 days ago
    Writing & ContentAuto-check passed

  • Tldr Prompt

    github/awesome-copilot

    Official

    Create tldr summaries for GitHub Copilot files (prompts, agents, instructions, collections), MCP servers, or documentation from URLs and queries.

    40k GitHub starsUsed in 1 repo~3.1k tokens
    Writing & ContentAuto-check passed

  • Cloud API Recipe Authoring

    TencentCloudBase/CloudBase-AI-Toolkit

    Author or revise a cloud-api-operations recipe (config/source/skills/cloud-api-operations/references/recipes/).

    1.1k GitHub stars~3.7k tokensUpdated today
    Agent WorkflowsAuto-check passed

  • Gearcoleco Romhacking

    drhelius/Gearcoleco

    Hack, modify, and translate ColecoVision and Super Game Module ROMs using the Gearcoleco emulator MCP server.

    142 GitHub stars~3.9k tokensUpdated 3 days ago
    Agent WorkflowsAuto-check passed

  • Weekly Project Digests

    thedotmack/claude-mem

    Turns a project's claude-mem timeline into a week-by-week narrative, splitting it by ISO week and running one subagent per week that receives the prior week's carry-forward block.

    99k GitHub stars~3.5k tokensUpdated yesterday
    Agent WorkflowsAuto-check passed

More from radimsem/remindb

All 11 skills in this repo

  • remindb Memory Writer

    radimsem/remindb

    Guides an agent writing to a remindb memory server: structured notes go in as files parsed into a node tree, single text facts go in through MemoryWrite.

    129 GitHub stars~2k tokensUpdated 2 mo ago
    Auto-check passed

  • remindb Read Path

    radimsem/remindb

    Covers the read-side tools of the remindb MCP server, a SQLite-backed agent memory, for orienting, searching, resyncing and tracing relations.

    129 GitHub stars~2.7k tokensUpdated 2 mo ago
    Auto-check passed

  • Add a remindb Bench Scenario

    radimsem/remindb

    Adds a new row to remindb's token-savings benchmark table by writing a scenario function that measures the naive shell-tool token cost against the remindb tool-call cost for the same task.

    129 GitHub stars~1.7k tokensUpdated 2 mo ago
    Auto-check passed

  • remindb Fuzz Target Author

    radimsem/remindb

    Adds a new Go fuzz target to remindb following its seed-corpus discipline: one target per logical surface, discovered automatically by its FuzzXxx function name.

    129 GitHub stars~1.7k tokensUpdated 2 mo ago
    Auto-check passed

  • remindb Integration Tests

    radimsem/remindb

    Explains how to add an end-to-end test scenario to remindb, choosing between a direct API test and an MCP test and using the shared helpers and fixtures.

    129 GitHub stars~1.8k tokensUpdated 2 mo ago
    Auto-check passed

  • Add an MCP Tool to remindb

    radimsem/remindb

    Checklist for adding a new Memory-prefixed tool to remindb's Go MCP server: tool file, registration, test, skill docs, plus the locking and logging rules.

    129 GitHub stars~2.1k tokensUpdated 2 mo ago
    Auto-check passed

Works with

Categories

Questions about Tune Temperature Policy

What does Tune Temperature Policy do?

A skill your agent uses when changing any field in pkg/temperature/Config (DecayRate, AccessBoost, ColdThreshold, NotifyThreshold, TickInterval) or modifying decayFactor / Score / cold-node…. Tune Temperature Policy is an agent skill from radimsem/remindb. Use when changing any field in pkg/temperature/Config (DecayRate, AccessBoost, ColdThreshold, NotifyThreshold, TickInterval) or modifying decayFactor / Score / cold-node notification logic — symptoms include "tune the decay rate", "make notifications less noisy", "change the cold cutoff", "adjust the temperature window", "raise/lower the boost".

When should I use Tune Temperature Policy?

Tune Temperature Policy fits situations like: changing any field in pkg/temperature/Config (DecayRate; notifyThreshold; modifying decayFactor / Score / cold-node notification logic — symptoms include tune the decay rate; make notifications less noisy.

How do I install Tune Temperature Policy in Claude Code?

Run `npx skills add radimsem/remindb --skill tune-temperature-policy -a claude-code`. Or copy the skill folder (.claude/skills/tune-temperature-policy in radimsem/remindb) into .claude/skills/tune-temperature-policy in your project. Claude Code loads it when a task matches its description.

How do I install Tune Temperature Policy in Codex?

Run `npx skills add radimsem/remindb --skill tune-temperature-policy -a codex`. Or copy the skill folder (.claude/skills/tune-temperature-policy in radimsem/remindb) into .agents/skills/tune-temperature-policy in your project. Codex loads it when a task matches its description.

Can I use Tune Temperature Policy 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 radimsem/remindb --skill tune-temperature-policy -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/tune-temperature-policy, .gemini/skills/tune-temperature-policy, .github/skills/tune-temperature-policy and .opencode/skills/tune-temperature-policy in your project.

What does Tune Temperature Policy need to run?

SKILL.md names no scripts, command-line tools or credentials: Tune Temperature Policy is instructions for the agent only.

Does Tune Temperature Policy 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 Tune Temperature Policy 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. Review the folder before installing.

What licence does Tune Temperature Policy use?

Tune Temperature Policy is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Tune Temperature Policy use?

About 2.3k tokens (SKILL.md is roughly 9.2k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Tune Temperature Policy?

Skills that share tags, products or a category with Tune Temperature Policy: Vss MCP Integration (open-edge-platform/edge-ai-libraries, 169 stars), Spark Report (breakstageaxe61/genspark-claw, 159 stars), Tldr Prompt (github/awesome-copilot, 40k stars) and Cloud API Recipe Authoring (TencentCloudBase/CloudBase-AI-Toolkit, 1.1k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Tune Temperature Policy?

radimsem (a GitHub user) maintains it in radimsem/remindb, which has 129 GitHub stars. The repository holds 11 skills in this directory. The repository was last updated on August 3, 2026.

Source: radimsem/remindb on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.