kirodotdev

ai-discover

Read-only discovery step of the auto-improvement loop. One bounded agent turn per cycle reads the focus files (Read/Grep/Glob only, no shell, no subagents) and replies with a JSON array of testable surfaces {file, line, symbol, rule, message, hypothesis}; the spine authors fixes and verifies them. Discovery only — no code changes and no verification happen here.

kirodotdev 4,973 1,058 Updated 1w ago
GitHub

Install

npx skillscat add kirodotdev/kirocrew/ai-discover

Install via the SkillsCat registry.

SKILL.md

ai-discover — the discovery step of the auto-improvement loop

This skill drives Phase A (discovery) of an auto-improvement cycle. It is
discovery only: it never applies a change, never runs the keep-or-revert A/B,
and never decides whether a finding is kept or drafted as a pull/merge request. Those
are the spine's deterministic Python gate / keeper / pipeline, which no model
can argue past. That separation is the point — the measurement is the product.

How discovery actually runs

The spine runs discovery as ONE bounded agent turn per cycle
(spine/agent_discovery.py discover_surfaces_via_agent), not a fan-out. Do
NOT spawn subagents: the turn must end with the JSON array itself, and children
spawned from an unattended runner orphan and hang the run (see the
SessionAgentRunner notes in spine/agent_runner.py).

  • Tools are Read, Grep and Glob only. There is no shell and no edit: discovery
    is read-only by capability, so you cannot run a profiler or write a test here.
  • Read the priority slice you are given, stop after 2-3 solid findings or the
    read budget, and reply with ONLY a JSON array (at most the stated limit) of
    {file, line, symbol, rule, message, hypothesis} — rule is AGENT, or
    COVERAGE for a coverage gap. Reply [] when nothing is testable. The final
    reply IS the output; there is no separate artifact file.
  • The same surface shape feeds both tracks. On the bug track the profile names
    the reproducing test path and author_bug_fix later writes the RED test and
    the fix in an isolated worktree; on the perf track author_perf_fix writes the
    edit. The spine dedups by (kind, basename::symbol) locus, then gates
    (RED -> GREEN -> STAYGREEN for bugs, serial pinned A/B for perf).

Reporting a candidate honestly

State the expected win as a hypothesis, not a result. The spine measures it;
if your estimate was wrong the candidate is reverted and that is a normal,
useful outcome. A guess presented as a measurement is the failure mode this
whole app exists to prevent.

What this skill never does

  • Never edits the ruler, the measurement harness, the tests-of-record, or
    anything outside the active target profile's edit allowlist. Those paths
    are mechanically rejected, so an edit there wastes the whole cycle.
  • Never publishes or merges a pull/merge request. Survivors are drafted by the
    spine as GitHub draft PRs or GitLab draft MRs, and a human publishes them.
  • Never fabricates a measured number. A fabricated win is the worst possible
    reward-hack, because it corrupts the record the loop reasons from.