Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
86 commits
Select commit Hold shift + click to select a range
20386f7
docs(plan): add Phase 0.9 first-run friction fixes, input/output cont…
mattleaverton Apr 1, 2026
6cd2824
wip: PR review workflow graph (v2)
mattleaverton Apr 1, 2026
159c0f4
wip: fix build-test script survival across branch switch
mattleaverton Apr 1, 2026
54a2b10
wip: PR review workflow v3 — investigate/decide split
mattleaverton Apr 2, 2026
855df38
Merge remote-tracking branch 'upstream/main'
mattleaverton Apr 2, 2026
1827e16
wip: make PR review workflow repo-agnostic
mattleaverton Apr 2, 2026
e8f4e50
docs(plan): platform reframe — layered architecture for software ops …
mattleaverton Apr 3, 2026
370230e
engine: export types and create L1/L2 package structure
mattleaverton Apr 3, 2026
9a4e7cd
engine: wire layered handler registration from cmd/kilroy/
mattleaverton Apr 3, 2026
ea8cb75
engine: add test proving tool-only graph works with L0-only registry
mattleaverton Apr 3, 2026
bd2885c
engine: rename codergen to agent throughout
mattleaverton Apr 3, 2026
2ae3885
style: gofmt formatting pass after rename
mattleaverton Apr 3, 2026
2899f32
rundb: add SQLite run database with migration runner
mattleaverton Apr 3, 2026
a93ec55
engine: wire RunDB into engine lifecycle events
mattleaverton Apr 3, 2026
12e565b
cli: update runs list/prune/status --latest to use RunDB
mattleaverton Apr 3, 2026
f1291cc
engine: add input contract system
mattleaverton Apr 3, 2026
9754cad
engine: add workspace abstraction
mattleaverton Apr 3, 2026
75d55fb
engine: add output contract
mattleaverton Apr 3, 2026
53e314a
server: add /runs API endpoints and /pipelines backward compat
mattleaverton Apr 3, 2026
3c178d3
cli: add --label flag, --older-than duration for prune
mattleaverton Apr 3, 2026
0ec438e
agents: add tmux session manager and invocation templates
mattleaverton Apr 4, 2026
ca173df
agents/tmux: add smoke tests for real CLI tools
mattleaverton Apr 4, 2026
71dae76
agents: add TmuxAgentHandler for CLI tool execution via tmux
mattleaverton Apr 4, 2026
0157a2c
agents: wire TmuxAgentHandler end-to-end and prove it works
mattleaverton Apr 4, 2026
bcc6052
docs(plan): update Phase 3 with Fabro learnings and implementation notes
mattleaverton Apr 4, 2026
c9d3113
engine: add GitOps interface for version control abstraction
mattleaverton Apr 4, 2026
b972b2b
workflows: implement GitHook using gitutil
mattleaverton Apr 4, 2026
c890b9d
engine: replace all gitutil imports with GitOps interface
mattleaverton Apr 4, 2026
f976c0a
cmd: auto-detect git mode and fix workspace git detection
mattleaverton Apr 4, 2026
f66d912
engine: add AutoDetectGitOps for backward-compatible git detection
mattleaverton Apr 4, 2026
0387940
workflows: extract HumanGateHandler to Layer 2
mattleaverton Apr 4, 2026
1a0224c
workflows: add workflow package support
mattleaverton Apr 4, 2026
2f74d68
workflows: add supervisor prototype with run health assessment
mattleaverton Apr 4, 2026
d554204
engine: document CXDB integration boundary for future extraction
mattleaverton Apr 4, 2026
c1b6976
engine: add canonical RunEvent envelope for all progress events
mattleaverton Apr 4, 2026
21e4288
agents: fix codex and opencode tmux templates
mattleaverton Apr 4, 2026
16da5d3
docs: add Phase 1-3 integration verification report
mattleaverton Apr 4, 2026
d4bb68e
docs: update verification report with codex success
mattleaverton Apr 4, 2026
e9e1237
workflows: add build-test workflow package (Phase 4.3)
mattleaverton Apr 6, 2026
772c4c6
workflows: add PR review workflow package (Phase 4.1)
mattleaverton Apr 6, 2026
286c385
server: add DB fallback for run detail, workflows endpoint, sort filter
mattleaverton Apr 7, 2026
a3e6edf
server: fix JSON casing, reconcile stale runs on startup
mattleaverton Apr 7, 2026
d55056b
server: add output download, provider recording, omitempty cleanup
mattleaverton Apr 7, 2026
403ac2c
server: add node turns endpoint, edge conditions, output download
mattleaverton Apr 7, 2026
eed251a
server: POST /runs accepts workflow packages, inputs, workspace
mattleaverton Apr 7, 2026
cb2caa7
server: fix run detail to use DB for completed runs in registry
mattleaverton Apr 7, 2026
3e1a830
docs(plan): file conventions, git diff tracking, and run observability
mattleaverton Apr 7, 2026
4c34ed0
engine: add .kilroy/ file conventions for inter-node data exchange
mattleaverton Apr 7, 2026
f156c70
engine: add output contract enforcement at graph and node levels
mattleaverton Apr 7, 2026
e01bf09
engine: add per-node git diff tracking
mattleaverton Apr 7, 2026
b4fc79d
server: add diff API endpoint for per-node git diffs
mattleaverton Apr 7, 2026
16462ec
server: add file browser API for run logs and workspace
mattleaverton Apr 7, 2026
bf49cb1
test: add DiffStat to testGitOps helper
mattleaverton Apr 7, 2026
e2e9413
docs(plan): canonical run log — single verbose activity log per run
mattleaverton Apr 7, 2026
7ea424c
engine: add RunLog type for canonical per-run activity logging
mattleaverton Apr 7, 2026
22c3f41
engine: wire RunLog lifecycle events at run/node/edge/checkpoint points
mattleaverton Apr 7, 2026
1d0fc8d
engine: stream tool stdout/stderr line-by-line through RunLog
mattleaverton Apr 7, 2026
c7fb5fc
server: add GET /runs/{id}/log endpoint with filtering and SSE streaming
mattleaverton Apr 7, 2026
6cf190e
agents: add conversation log parsers for Claude, Codex, and OpenCode
mattleaverton Apr 7, 2026
7b69581
engine: add git activity events to RunLog (worktree.created, commit)
mattleaverton Apr 7, 2026
fd97407
fix: provider recording timing and model passing to CLI tools
mattleaverton Apr 7, 2026
ff668ae
workflows: add multi-tool exercise for observability testing
mattleaverton Apr 7, 2026
02d5043
agents: isolate CLI tool auth for headless operation
mattleaverton Apr 8, 2026
2600edb
engine: add --skip-preflight flag for tmux-managed sessions
mattleaverton Apr 8, 2026
24034fe
fix: CLI tool model IDs and auth format from first exercise run
mattleaverton Apr 8, 2026
a09f2b6
fix: codex auth.json uses OPENAI_API_KEY field, not token
mattleaverton Apr 8, 2026
9204aa9
workflows: switch codex model to codex-mini-latest
mattleaverton Apr 8, 2026
3503cfa
fix: codex template uses sandbox flag directly, gpt-5.4-nano model
mattleaverton Apr 8, 2026
64b9d75
agents: capture structured JSONL output for agent log parsing
mattleaverton Apr 8, 2026
b207c72
fix: disable codex web_search to support nano/mini models
mattleaverton Apr 8, 2026
c4e16c4
fix: rewrite codex and opencode log parsers for actual JSONL formats
mattleaverton Apr 8, 2026
a9c7d13
fix: pr-review workflow for worktree and tmux agent execution
mattleaverton Apr 8, 2026
7f65dfb
agents: real-time agent log streaming during execution
mattleaverton Apr 8, 2026
e722916
fix: bump stall watchdog from agent log tailer
mattleaverton Apr 8, 2026
0f82b1b
fix: record run failure in DB on engine error return
mattleaverton Apr 8, 2026
1de1253
server: SQLite concurrent writes, invocation capture, prefix IDs, can…
mattleaverton Apr 10, 2026
ee21f60
server: embed and serve dashboard UI at /ui/
mattleaverton Apr 13, 2026
c498d05
engine: loop primitive, per-iteration artifact capture, label filters
mattleaverton Apr 13, 2026
152bf92
engine: per-iteration attempt numbering, script capture, loop UI
mattleaverton Apr 13, 2026
d8073ce
engine: concurrent split/join primitive
mattleaverton Apr 13, 2026
c39598c
engine+validate: kill subprocess group on cancel, reject nested concu…
mattleaverton Apr 13, 2026
ac28ea6
cli+workflows: quick-launch package + runs show command
mattleaverton Apr 14, 2026
942ab31
skills: quick-launch skill + using-kilroy refresh
mattleaverton Apr 14, 2026
da2e759
skills/quick-launch: slash command + stable paths
mattleaverton Apr 14, 2026
09533fe
quick-launch: ergonomics pass + workspace fix + install script
mattleaverton Apr 14, 2026
c0fcd1f
engine: isolate codex runs from user's shell profile
mattleaverton Apr 15, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 27 additions & 0 deletions AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -131,6 +131,33 @@ Runs live under `~/.local/state/kilroy/attractor/runs/<run_id>/`. Key files:
- `progress.ndjson` — full event log (stage starts/ends, edge selections, LLM retries).
- `manifest.json` — run metadata (goal, graph, repo, base SHA).

### Agent Backend Configuration

Agent nodes (`shape=box`, `agent_tool="claude"`) require specific backend and handler configuration for proper agent log capture:

- **`backend: cli`** in the run config — invokes the actual CLI binary (`claude`, `codex`, `opencode`) with `--output-format stream-json`, producing `agent_output.jsonl` with full conversation logs (tool calls, thinking, responses). The server parses this into structured agent events for the UI.
- **`backend: api`** — uses the Anthropic HTTP API directly. Produces `events.ndjson` in a different format. The server does NOT currently parse this into UI-visible agent events. Use `backend: cli` for runs where you want the UI to show agent conversation detail.
- **`--tmux` flag** — required for agent nodes that use CLI backends. Registers `TmuxAgentHandler` which runs agent CLIs in tmux sessions for reliable headless execution. Without `--tmux`, the default `AgentHandler` is used (API-only path).
- **`--package` flag** — points to a workflow package directory (e.g., `workflows/pr-review/`). Copies scripts, prompts, and graph into the worktree at `.kilroy/package/`.

Example production PR review launch:
```bash
./kilroy attractor run --detach --tmux \
--package workflows/pr-review \
--config run.yaml \
--no-cxdb --skip-cli-headless-warning \
--input '{"pr_repo": "owner/repo", "pr_number": 123}'
```

The run config must specify `backend: cli` for providers used by agent nodes:
```yaml
llm:
cli_profile: real
providers:
anthropic:
backend: cli
```

### PR Review Process

For PRs we want to accept: check out the PR branch into a worktree, review, add fix-up commits, then non-squash merge — this preserves contributor credit while maintaining code quality.
Expand Down
Loading
Loading