agentleFS
lbusselllbussell/agent-skills/skills/dream/SKILL.md

dream

Skill0 starsChanged 5 days agoMITView on GitHub

Audit recent cross-repo sessions to propose additions, corrections, demotions, and prunes across instructions and skills. Use for /dream, nightly session reviews, or consolidating lessons across recent chats.

44 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use ingest, resume and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • JaimeGBASkill

    ingest

    Digest an external source (article, paper, transcript, URL, book chapter) and integrate its content into the knowledge base

    Also in 4 projects that use this

  • genomewalkerSkill

    resume

    Resume a thread by loading its ~800-token context capsule

    Also in 3 projects that use this

  • MicrosoftSkill

    init

    Also in 4 projects that use this

  • Hugging FaceSkill

    simplify

    Refactor code after tests pass. The "Refactor" phase of Red-Green-Refactor.

    Also in 3 projects that use this

  • SentrySkill

    deslop

    Strip AI-generated noise from the diff before pushing — extra comments a human wouldn't write, defensive try/catch in trusted code paths, casts to any, inline imports in Python, and other style inconsistencies with the surrounding file. Use this immediately before commit so the diff stays clean.

    Also in 3 projects that use this

  • AnthropicSkill

    start

    Set up your bio-research environment and explore available tools. Use when first getting oriented with the plugin, checking which literature, drug-discovery, or visualization MCP servers are connected, or surveying available analysis skills before starting a new project.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 2 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 43 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use the time window from the prompt, or default to the last 24 hours.

    line 12
  2. Very specific:

    Check recent dream sessions and child sessions so you skip proposals the user already archived or declined.

    line 15
  3. Distinctive:

    If no substantive sessions remain, report that nothing needed review and stop.

    line 16
  4. Very specific:

    Read small groups (15 total turns or fewer) directly.

    line 21
  5. Very specific:

    Read ~/.copilot/dream-watchlist.md (if present) and each candidate target file before drafting proposals:

    line 30
  6. Very specific:

    Update ~/.copilot/dream-watchlist.md per references/rubric.md to prune expired entries, remove promoted matches, and record validated …

    line 39
  7. Very specific:

    Do not edit ~/.copilot/copilot-instructions.md or ~/.agents/skills/ before explicit user approval.

    line 42
  8. Very specific:

    Call list_sessions_and_chats and post a digest covering child session links (app_url when present), any ~/.copilot diff, watchlist changes, …

    line 43
  9. Very specific:

    When the user approves a ~/.copilot/copilot-instructions.md or ~/.agents/skills/ diff in chat, apply that edit in place.

    line 47

“Common” is measured: kept by at least 6 in 10 of the 43 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Dream

Consolidate lessons from recent sessions into reviewable updates across user and repository instructions and skills.

1. Collect sessions

  1. Use the time window from the prompt, or default to the last 24 hours.
  2. Collect candidate sessions from session_store_sql (cloud and local) and list_sessions_and_chats using references/rubric.md.
  3. Exclude the current session, previous dream runs, no-op automations, and single-turn lookups.
  4. Check recent dream sessions and child sessions so you skip proposals the user already archived or declined.
  5. If no substantive sessions remain, report that nothing needed review and stop.

2. Mine transcripts

  1. Group sessions by repository or cwd and map configured repositories with list_repositories.
  2. Read small groups (15 total turns or fewer) directly; for larger groups, spawn one read-only general-purpose subagent per repository group so it can call session_store_sql.
  3. Extract four signal types with session IDs and turn numbers:
    • User corrections and repeated agent mistakes.
    • Commands, flags, workflows, or preferences that succeeded after trial and error.
    • Existing instructions or skills that misled the agent, drifted from the code, or went ignored.
    • Visible skills that failed to trigger or manual steps that repeated across turns.

3. Audit targets and filter

  1. Read ~/.copilot/dream-watchlist.md (if present) and each candidate target file before drafting proposals:
    • User instructions: ~/.copilot/copilot-instructions.md.
    • User skills: ~/.agents/skills/ and agent-skills (main_repo_path from list_repositories, or local main when running in an agent-skills worktree).
    • Repository instructions: <repo>/AGENTS.md and <repo>/.github/copilot-instructions.md.
    • Repository skills: <repo>/.github/skills/ and <repo>/.agents/skills/.
  2. Match signals against ~/.copilot/dream-watchlist.md, then route and filter candidates with references/rubric.md.

4. Stage proposals

  1. Update ~/.copilot/dream-watchlist.md per references/rubric.md to prune expired entries, remove promoted matches, and record validated single-session candidates.
  2. For git-backed targets (each affected repository and agent-skills), spawn one child worktree session per repository with create_session using references/rubric.md.
  3. End your turn while child sessions run; when idle notifications arrive, archive any child with zero changes and keep every child with a diff.
  4. Do not edit ~/.copilot/copilot-instructions.md or ~/.agents/skills/ before explicit user approval; render unified diff blocks for them in chat.
  5. Call list_sessions_and_chats and post a digest covering child session links (app_url when present), any ~/.copilot diff, watchlist changes, and structural check candidates.

5. Apply non-git proposals

  1. When the user approves a ~/.copilot/copilot-instructions.md or ~/.agents/skills/ diff in chat, apply that edit in place.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.