ModLens Image Vision Bridge
liustack/modlens
Gives text-only models sight by running the modlens CLI on an image path or URL and returning structured JSON evidence with transcribed text, layout and semantics.
Premium judgment; cheap repository specialists; deterministic helpers for objective integrity.
MITAuto-check passed
Install Deepseek And Destroy
$ npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install frozenpepper/deepseek-and-destroy deepseek-and-destroy --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
Claude Code skills documentation · loads skills from .claude/skills/
Install the "deepseek-and-destroy" agent skill from https://github.com/frozenpepper/deepseek-and-destroy/tree/main into .claude/skills/deepseek-and-destroy/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "deepseek-and-destroy", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install frozenpepper/deepseek-and-destroy deepseek-and-destroy --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "deepseek-and-destroy" agent skill from https://github.com/frozenpepper/deepseek-and-destroy/tree/main into .agents/skills/deepseek-and-destroy/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "deepseek-and-destroy", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install frozenpepper/deepseek-and-destroy deepseek-and-destroy --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "deepseek-and-destroy" agent skill from https://github.com/frozenpepper/deepseek-and-destroy/tree/main into .cursor/skills/deepseek-and-destroy/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "deepseek-and-destroy", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install frozenpepper/deepseek-and-destroy deepseek-and-destroy --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "deepseek-and-destroy" agent skill from https://github.com/frozenpepper/deepseek-and-destroy/tree/main into .gemini/skills/deepseek-and-destroy/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "deepseek-and-destroy", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install frozenpepper/deepseek-and-destroy deepseek-and-destroyInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "deepseek-and-destroy" agent skill from https://github.com/frozenpepper/deepseek-and-destroy/tree/main into .github/skills/deepseek-and-destroy/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "deepseek-and-destroy", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install frozenpepper/deepseek-and-destroy deepseek-and-destroy --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "deepseek-and-destroy" agent skill from https://github.com/frozenpepper/deepseek-and-destroy/tree/main into .opencode/skills/deepseek-and-destroy/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "deepseek-and-destroy", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
deepseek-and-destroy- GitHub stars
- 152
- Token cost
- ~1.9k tokens
- SKILL.md length
- 814 words
- Files
- 68 (incl. scripts)
- Skills in repo
- 10
- Repo updated
- First seen
- Licence
- MIT
At a glance
Premium judgment; cheap repository specialists; deterministic helpers for objective integrity.
- SKILL.md covers Ownership and trust, Premium discipline, Context locality and resume and Normal execution, plus 1 more section
What it does
Deepseek And Destroy is an agent skill from frozenpepper/deepseek-and-destroy. Premium judgment; cheap repository specialists; deterministic helpers for objective integrity.
Its SKILL.md is about 1.9k tokens, which your agent loads only when the skill is triggered. The skill folder holds 70 other files, including scripts (for example `CHANGELOG.md`, `CLAUDE.md` and `CODEX.md`). Compatibility notes: codex, claude-code, opencode, kilo, and comparable coding harnesses
It works with DeepSeek. The licence is MIT.
Example prompts
- “/deepseek-and-destroy”
Requirements
- Python 3
- Compatibility (from SKILL.md): codex, claude-code, opencode, kilo, and comparable coding harnesses
What it can do on your machine
Read from SKILL.md and the folder at commit ab72d2b. 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.
Compatibility
codex, claude-code, opencode, kilo, and comparable coding harnesses
From compatibility in the SKILL.md frontmatter.
Context cost
Deepseek And Destroy loads about 1.9k tokens when it runs. Until then it costs about 29 tokens; SKILL.md has 814 words of instructions outside code blocks.
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 frozenpepper/deepseek-and-destroy at commit ab72d2b, republished under its MIT licence (© frozenpepper). 814 words, ~1,860 tokens.
Download SKILL.mdSave it as
.claude/skills/deepseek-and-destroy/SKILL.md (or your agent's skills folder). This skill also uses 67 other files; get the full folder from GitHub.- name
- deepseek-and-destroy
- description
- Premium judgment; cheap repository specialists; deterministic helpers for objective integrity.
- compatibility
- codex, claude-code, opencode, kilo, and comparable coding harnesses
- license
- MIT
- metadata.default-worker-harness
- opencode-cli
- metadata.default-worker-model
- opencode-go/deepseek-v4-flash
- metadata.workspace-root
- DeepSeekAndDestroy
DeepSeek and Destroy
Spend premium context only where premium judgment is required.
Continue the authoritative plan until COMPLETED, HUMAN-BLOCKED, PAUSED-BY-USER, or ABANDONED. Tasks/reviews/retries/phases/compaction are not terminal.
Ownership and trust
- Parent: authority, decomposition, role choice, consequential decisions, acceptance, phase approval.
- Workers: repository-scale discovery, implementation, repair, review, verification, recovery, phase audit.
- Python: objective facts only—immutable bindings/hashes, lifecycle, source movement, explicit authority restrictions, and resume continuity. Never infer engineering meaning from prose.
- Clerk: optional cheap semantic compressor; interpret existing evidence only. Never invent/verify missing proof, edit project state, waive integrity, accept work, or recurse.
Trust the specialist chain. After mutation, fresh Reviewer is routine technical verification; parent does not repeat it for reassurance. Parent self-verification belongs at the frozen phase gate or an explicit authority/judgment escalation.
Premium discipline
Use smallest sufficient evidence: mechanics → bounded --surface → Clerk → targeted evidence → full report. Stop when supported.
- Do not consume Implementer/Fixer output to duplicate specialist judgment. Recovery/lifecycle/contract shaping may inspect bounded evidence to recover findings/pointers without re-reviewing.
- Do not duplicate specialist investigation/verification. A bounded source read/search may cheaply answer a contract-shaping/parent-only question; delegate broad tracing, proof, tests, and re-review.
- Contracts are deltas over reviewed authority: name exact steps/sections and only unstated requirements; add write bounds only when authority defines them.
- Do not hand-edit accepted worker/project artifacts. Route corrections as a bounded new revision/task.
- If DSD mechanics fail, delegate bounded framework investigation when possible; premium source archaeology is a last resort.
- Routine execution is silent. Do not yield merely to summarize: execute safe
next_actionfirst. Speak only for completion, real decision/block/escalation, reviewed phase-end, user pause/request, or material in-flight finding. When speaking, assume the user saw none of the worker output: give the current objective, material status and why it matters, recommendation/decision if any, and next action. Be concise; avoid unexplained internal shorthand.
Context locality and resume
Parent hot context is this file + one harness adapter. Cold-load: WORKSPACE.md for lifecycle/recovery; OPENCODE.md for transport/session trouble; COMPACTION.md for checkpoint recovery; PROMPTS.md for task/handoff authoring/debugging. On the first abnormal lifecycle symptom load WORKSPACE.md before improvising; on the first transport/provider/session symptom load OPENCODE.md before diagnosing or retrying. Workers get run facts + worker/COMMON.md + one role + task; proof recipes only when requested.
On a fresh parent session, do not reconstruct the run. Resolve identity from explicit binding or minimal DSD state.json metadata—not plans/git/reports/session history—then read live state first. Use state.run_root verbatim. Execute a mechanical next_action immediately; for semantic action read only named decision/evidence/authority. If active runs stay ambiguous, require exact run authority.
Show full SKILL.md (407 more words)Show less
Normal execution
Once the run is known, read authority for the next parent decision and delegate repository-scale measurement. One task = one independently reviewable semantic objective. Sweep a recognizable same-root-cause family once inside that boundary. If writer orientation needs substantial discovery, measure first with read-only Discovery and pass its map forward. Implementer/Fixer choose the files needed to satisfy authority. Do not predict the diff; supply known entry points, symbols, invocations, findings, and evidence paths as orientation. write_paths is only for authority-imposed hard bounds; Python enforces it.
Repeated run-specific instructions belong in the next worker-rules revision; universal doctrine stays in the skill and task authority in the contract. Run rules may narrow, never contradict protocol or reintroduce report grammar.
Workers own routine engineering choices. If one uncovers a consequential decision beyond current authority, consume its bounded DECISION_REQUIRED, record the parent decision, and resume the same role/session with it as exact input. Recut the contract only if authority, scope, or acceptance materially changed.
dsd_attempt.py launch --run-root … --phase-id … --task-id … --role … [--detach]
dsd_attempt.py wait --run-root … --phase-id … --task-id … # detached only
dsd_attempt.py gate --run-root … --phase-id … --task-id … [--surface]Use --surface only at a parent semantic boundary; intermediate gates return mechanics only. Wait quiescently: one timeout without terminal evidence is a non-event. Repeated timeouts plus a credible stall signal may trigger one bounded diagnosis; log age/size or recorded liveness are clues, never proof, and continuous model-visible polling stays forbidden. Role changes use fresh contexts. A terminal exit, including 0, ends only that process turn; unfinished trustworthy same-role work should resume its recorded OpenCode session in a new attempt. See OPENCODE.md.
Implementer → fresh Reviewer
│
FAIL └→ fresh Fixer → fresh Reviewer …
↓
parent decision boundary
↓ only if useful
ClerkMissing technical proof goes to targeted Verification/Review, never Clerk or report-format rerun. Load WORKSPACE.md only for abnormal lifecycle handling.
Integrity, phase, human boundary
A clean mechanical gate means safe to interpret, never semantic PASS. Hard failures are objective integrity failures only. Worker prose is a claim/evidence index, not proof of artifact state; parent decides semantics and accept-task records provenance.
Phase close: finish writers → exercise finalization operations that establish/refer to the final snapshot → freeze → required post-freeze Verification → fresh Phase Auditor → parent phase judgment. Finalization must not require later mutation inside that snapshot or a self-invalidating dependency cycle; later mutation invalidates stale phase evidence. After approved close, perform adapter phase cleanup (OPENCODE.md).
Scope-observed mutation is exclusive per checkout: read-only attempts overlap only each other; no worker/parent may mutate observed project state while one is live. Parent project edits count. Writer + read-only requires isolated worktrees. Ask the human only for uninferable authority, access/authorization, destructive/paid/live permission, persistent worker unavailability, or irreconcilable authority conflict. Give finding and recommendation, not a menu.
© frozenpepper, MIT. 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 67 other files (scripts) in the repository root of frozenpepper/deepseek-and-destroy.
- SKILL.md
- .gitignore
- CHANGELOG.md
- CLAUDE.md
- CODEX.md
- COMPACTION.md
- CONFIG.example.md
- HARNESS.md
- KILO.md
- LICENSE
- OPENCODE.md
- PROMPTS.md
- README.md
- WORKSPACE.md
- adapters/README.md
- adapters/claude/settings.fragment.json
- adapters/codex/config.fragment.toml
- adapters/codex/hooks.json
- … and 50 more
Open the folder on GitHubat commit ab72d2b
Compare with similar skills
Deepseek And Destroy 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.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Deepseek And Destroy this skillfrozenpepper/deepseek-and-destroy | 152 | — | ~1.9k | Automated safety check: Pass | MIT | |
| ModLens Image Vision Bridgeliustack/modlens | 4.2k | — | ~1.3k | Automated safety check: Notes | MIT | |
| Distilly Person Profile Buildertitanwings/distilly | 25k | — | ~15k | Automated safety check: Notes | MIT | |
| Evals Contextzgsm-ai/costrict | 4.4k | 1 repos | ~1.9k | Automated safety check: Pass | Apache-2.0 | |
| J SpaceTiger3807861189/J-Space-Cognition-Suite | 3k | — | ~3k | Automated safety check: Pass | Apache-2.0 | |
| Weavebench Cua ReproduceAMAP-ML/LongHorizon-Harness | 1.7k | — | ~1.6k | Automated safety check: Pass | MIT |
Similar skills
Distilly Person Profile Builder
titanwings/distilly
Distills source material about a colleague, a relationship or a celebrity into reusable Person Profiles that an agent can later work from, in English or Chinese.
Evals Context
zgsm-ai/costrict
Provides context about the CoStrict evals system structure in this monorepo.
J Space
Tiger3807861189/J-Space-Cognition-Suite
Operate a selective workspace for complex reasoning, long tasks, repository engineering, coordinated agents, and authorized security analysis.
Weavebench Cua Reproduce
AMAP-ML/LongHorizon-Harness
Reproduce CUA-Harness experiments on WeaveBench from a GitHub checkout.
Vision Skills
Anionex/agent-vision-toolkit
Local vision CLIs: glance (describe/ask/OCR an image), ground (locate a target, pixel box), detect (element inventory), trace (image to SVG geometry), crop (cut a pixel box to a file), and…
More from frozenpepper/deepseek-and-destroy
All 10 skills in this repoDsd Evidence Clerk
frozenpepper/deepseek-and-destroy
Cheap read-only semantic interpreter/compressor for existing DSD evidence.
Dsd Fixer
frozenpepper/deepseek-and-destroy
Repair supplied findings for one bounded DSD task. An agent skill from frozenpepper/deepseek-and-destroy.
Dsd Implementer
frozenpepper/deepseek-and-destroy
Implement one bounded DSD task end-to-end without self-approval.
Dsd Phase Auditor
frozenpepper/deepseek-and-destroy
Fresh read-only audit of one frozen DSD phase. An agent skill from frozenpepper/deepseek-and-destroy.
Dsd Reviewer
frozenpepper/deepseek-and-destroy
Fresh adversarial reviewer for one bounded DSD task. An agent skill from frozenpepper/deepseek-and-destroy.
Dsd Verification
frozenpepper/deepseek-and-destroy
Verify one explicitly assigned technical predicate; read-only unless exact artifact writes are contract-authorized.
Related
Works with
Questions about Deepseek And Destroy
What does Deepseek And Destroy do?
Premium judgment; cheap repository specialists; deterministic helpers for objective integrity. Deepseek And Destroy is an agent skill from frozenpepper/deepseek-and-destroy. Premium judgment; cheap repository specialists; deterministic helpers for objective integrity.
How do I install Deepseek And Destroy in Claude Code?
Run `npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a claude-code`. Or copy the skill folder (the frozenpepper/deepseek-and-destroy repository) into .claude/skills/deepseek-and-destroy in your project. Claude Code loads it when a task matches its description.
How do I install Deepseek And Destroy in Codex?
Run `npx skills add frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a codex`. Or copy the skill folder (the frozenpepper/deepseek-and-destroy repository) into .agents/skills/deepseek-and-destroy in your project. Codex loads it when a task matches its description.
Can I use Deepseek And Destroy 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 frozenpepper/deepseek-and-destroy --skill deepseek-and-destroy -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/deepseek-and-destroy, .gemini/skills/deepseek-and-destroy, .github/skills/deepseek-and-destroy and .opencode/skills/deepseek-and-destroy in your project.
What does Deepseek And Destroy need to run?
SKILL.md names no scripts, command-line tools or credentials: Deepseek And Destroy is instructions for the agent only. Our summary lists: Python 3. Compatibility (from SKILL.md): codex, claude-code, opencode, kilo, and comparable coding harnesses.
Does Deepseek And Destroy 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 Deepseek And Destroy 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 Deepseek And Destroy use?
Deepseek And Destroy is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Deepseek And Destroy use?
About 1.9k tokens (SKILL.md is roughly 7.4k 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 Deepseek And Destroy?
Skills that share tags, products or a category with Deepseek And Destroy: ModLens Image Vision Bridge (liustack/modlens, 4.2k stars), Distilly Person Profile Builder (titanwings/distilly, 25k stars), Evals Context (zgsm-ai/costrict, 4.4k stars) and J Space (Tiger3807861189/J-Space-Cognition-Suite, 3k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Deepseek And Destroy?
frozenpepper (a GitHub user) maintains it in frozenpepper/deepseek-and-destroy, which has 152 GitHub stars. The repository holds 10 skills in this directory. The repository was last updated on August 22, 2026.
Source: frozenpepper/deepseek-and-destroy on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.