Skip to content
cccskills
FreePublished on GitHub

instinct-export

Use when the user wants to share or back up learned instincts — validates the local instincts file, allowlists the six known fields, strips secrets and home-directory paths, and writes a sanitized JSON artifact the user names. Never modifies the instincts file itself.

The English summary is being prepared. Showing the original description.

See how to install

Files included (1)

  • SKILL.md3.9 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

instinct-export — sanitize and share learned instincts

Concept inspired by the instinct/skill-create ideas in affaan-m/ECC; no code copied.

Instincts are small learned behaviours stored locally under ~/.aos/instincts/instincts.json (format defined in /instinct-status). Exporting one means making it portable — into version control, a teammate's machine, a dotfile repo. That is exactly where leaks happen, so sanitization is the job, not a nicety.

1. Read and validate the source

Read ~/.aos/instincts/instincts.json. If it is missing, the file is not silently treated as empty: report that there is nothing to export and stop. If the JSON does not parse, fail loudly with the parse error — never export garbage "with best effort".

Shared bounds, same as on the import side: trigger and action must be JSON strings ≤ 200 / ≤ 400 characters, no newlines or control characters; at most 200 instinct entries. A store that fails these bounds is corrupt, not exportable — fail loudly with the reason the same way as the parse-error path above, never ship an artifact /instinct-import would reject wholesale.

Field allowlist: every exported entry keeps exactly the six shared fields — id, trigger, action, confidence (0–1), source, updated — and every other field is dropped. No auxiliary metadata travels with the entry. Nothing on the allowlist is negotiable; extra fields are removed, not published.

2. Redact — drop anything that looks like a credential or a leak

For every entry, scrub trigger and action against these redaction rules (a hit means remove the content, or the whole entry if the lesson dies with it):

  • sk-… prefixed keys.
  • ghp_… tokens.
  • github_pat_… tokens.
  • xox-prefixed Slack tokens (xoxp-, xoxb-, …).
  • AWS AKIA… access-key ids.
  • JWTs: three dot-separated base64 parts (header.payload.signature).
  • Any base64 or hex run of 32+ characters.
  • URL userinfo: user:pass@… in any URL.
  • E-mail addresses.
  • Absolute home paths: /Users/<name>/…, /home/<name>/…, C:\Users\<name>\…. Rewrite to ~/… or describe the location in words. Relative paths (src/lib/) are safe.

An instinct that only exists to memorialize a secret is dropped, not partially redacted. A missing entry is recoverable; a leaked credential is not.

3. Re-scan the output before writing

After redaction and field-filtering, serialize the JSON and scan the whole serialized output once more against the section-2 list before any write. Redaction that happens only on the entry objects can miss paths, e-mails or tokens introduced by formatting or by fields added late; the final scan is what closes that gap. If the re-scan still finds a hit, fix it and re-scan again — never write with a known hit remaining.

4. Write the artifact

Default if the user names no path: a scratch/output folder the user names — for example ~/instinct-export/instincts-YYYY-MM-DD.json. Never the home root (~/instincts-export-….json is out), never a dotfiles repo path, never inside ~/.aos/. If the user names a path inside the AOS state dir, refuse it — resolve .. segments and symlinks first and check the resolved path, not the spelled-out one. Never overwrite an existing export file without the user's explicit yes. The export keeps version: 1 and an exported: <ISO date> key at the top level; the instincts file itself is never modified.

5. Report

Summarize: N instincts exported, K sanitized (list the ids), M omitted and why (one line each). Show the output path and state that the original file was untouched.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

See all skills by hybridlabor-api

Report a problem with this skill