Runs before the agent uses a tool for Bash, Read and Grep tool calls, executing snapshot-pre-bash.sh, check-readme.sh, guard-git.sh, guard-pr-body.sh, commitlint-local.sh, pre-commit.sh, lint-staged.sh and enrich-context.sh with --show-toplevel (9 commands). From optave/ops-codegraph-tool.
Runs after a tool call finishes for Edit, Write and Bash tool calls, executing update-graph.sh, track-edits.sh, track-bash-writes.sh, track-moves.sh and post-git-ops.sh with --show-toplevel (5 commands). From optave/ops-codegraph-tool.
92 3d agoA
tokens not measured
copy · 94%Apache-2.0
Run a full dogfooding session against a codegraph release (stable from npm, or dev from GitHub release tarballs) — install, test all commands, compare engines, find bugs, and write a report.
Solve open issues end-to-end in batches of N — one fresh branch per issue, PR, review convergence, and merge — looping across batches until the whole backlog for --author is drained (or a safety cap is hit), then draining any stragglers in a final sweep phase. Never operates outside the current repo; if there is…
Human-gated, single-issue delivery. Preflights the session config (Opus, effort >= xhigh, auto-accept), then runs PLAN -> adversarial critic -> plan-sweep and STOPS for a human to approve the plan via a checkbox on the [Plan] PR. Re-invoked on the plan PR (after the box is ticked) it verifies non-forgeable approval…
Audit WASM/native engine correctness parity across all resolution fixtures and fix any divergence at the root cause — both engines must produce identical graphs.
Resolve merge conflicts on a PR with full context awareness — identifies which PR introduced each conflicting change, understands both sides' intent, and resolves without silently dropping functionality.
Check all open PRs, resolve conflicts, update branches, address Claude and Greptile review concerns, fix CI failures, and retrigger reviewers until clean.
Audit codebase files against the 4-pillar quality manifesto using RECON work batches, with batch processing and context budget management (Titan Paradigm Phase 2).
Adopt extracted helpers — find dead symbols from forge, wire them into consumers, replace duplicated inline patterns, and gate on dead-symbol delta (Titan Paradigm Phase 4.5).
Map a codebase's dependency graph, identify hotspots, name logical domains, propose work batches, and produce a ranked priority queue for autonomous cleanup (Titan Paradigm Phase 1).
Run the full Titan Paradigm pipeline end-to-end by dispatching each phase to sub-agents with fresh context windows. Orchestrates recon → gauntlet → sync → forge → grind (+ repo-provided parity audit) automatically.