agentleFS
hazarsozerhazarsozer/ren-os/skills/suggestions/SKILL.md

suggestions

Skill2 starsChanged 2 months agoMITView on GitHub

Use when the friend wants to review pending suggestions: rare, high-stakes items a producer (retrospective, promotion, doctrine, wiki-health) has raised for explicit approve/reject — never auto-applied. Triggers on the /ren:suggestions slash command. The single interactive decide surface over lib.suggestions' durable store (Task 14); this is where an accepted decision actually becomes a write.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If it returns "No pending suggestions.", say so and stop — nothing else to do.

    line 66
  2. Distinctive:

    Ask the friend to Approve or Reject (AskUserQuestion, or plain chat if unavailable).

    line 73
  3. Distinctive:

    Never batch-apply: Never apply a suggestion without an explicit per-item answer from the friend.

    line 91

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

suggestions

The single interactive approve/reject surface over lib.suggestions' durable store (Task 14). Suggestions are RARE AND HIGH-STAKES by design (spec §1.2) — this is the only place a suggestion's payload actually gets applied. lib.suggestions.decide itself is a pure state transition; this skill's accept() is what turns an "accepted" decision into a real write.

When to use this skill

  • Friend invokes /ren:suggestions — review whatever's pending
  • A producer (retrospective, promotion, doctrine, wiki-health) just raised a new suggestion and the friend wants to act on it now rather than later

When NOT to use this skill

  • The friend wants to record something new mid-session → /ren:pin
  • The friend wants a coherence sweep, not a decision queue → /ren:wiki-health

Behavior

  1. Call skills.suggestions.lib.render_list().
  2. If it returns "No pending suggestions.", say so and stop — nothing else to do.
  3. Otherwise walk the pending suggestions ONE AT A TIME, in the order render_list() returned them (oldest first):
    • Show render_suggestion(s) for the current suggestion — title, rationale, producer, and (for page_write suggestions) a content preview.
    • Ask the friend to Approve or Reject (AskUserQuestion, or plain chat if unavailable). Get an explicit per-item answer — never assume, never infer from silence.
    • On approve: call skills.suggestions.lib.accept(sid, session). Report what happened from the returned {"sid", "applied", "detail"} — if applied is False, say so and show detail (e.g. a duplicate-content no-op, or the two pages + evidence for a review_contradiction suggestion the friend now reconciles conversationally). The result also carries "decision_recorded" (True/False) on every path that followed an apply attempt — absent only on the up-front already-decided check, so use .get("decision_recorded"). If it's False, the change may already have landed (applied can still be True) but the decision itself wasn't recorded — the suggestion stays pending and will be offered again; mention that to the friend rather than treating it as a clean accept.
    • On reject: call skills.suggestions.lib.decline(sid) — durable, this suggestion is never re-offered.
    • Move to the next suggestion.
  4. Never batch-apply. Never apply a suggestion without an explicit per-item answer from the friend.

Apply routes (by suggestion kind / payload action)

  • page_write → lib.memory.queue.propose then queue.approve_and_apply(qid, who="suggestions"). Instruction-plane (global/) pages go through the SAME human-gated door as ever — the recorded suggestion decision IS the human approval. If propose returns a synthetic noop-duplicate entry (content already matches the live page), accept() treats it as applied=False with a "content already on page" detail and still records the accept decision — never a half-applied state.
  • promote_to_global → lib.memory.promotion.promote_to_global(source_page, session) then approve_and_apply.
  • refresh_claude_md → lib.adapter.claude_md.write_global_claude_md().
  • review_contradiction → applies nothing; returns the two page paths and evidence so the session can reconcile them conversationally.
  • orphan_page (#55) → applies nothing; hands the orphan page off to the session to place (a hub, a map, log.md), same convention as review_contradiction.

Any apply failure is caught inside accept() and surfaced in detail — the decision stays recorded either way.

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.