Add self-correction loop: logs, call, and shot commands by molefrog · Pull Request #30 · molefrog/moi

molefrog · 2026-07-07T07:42:55Z

Implements the three-leg self-correction feedback loop for applets, allowing the agent to detect runtime failures, test server functions, and see what applets actually look like without waiting for user feedback.

Summary

Adds three new CLI commands (moi logs, moi call, moi shot) that close the gap between successful builds and actual runtime behavior. The agent can now:

Feel (moi logs) — query an in-memory error journal of applet runtime failures
Poke (moi call) — invoke server functions directly through the same worker pool as browser RPC
See (moi shot) — render applets to PNG through a live workspace tab

Key Changes

Server-side error journal (server/applet-log.ts)

Per-workspace, in-memory ring buffer (100 entries max) tracking applet failures
Records five error sources: build, load, render, window, rpc
Deduplicates identical errors (same source + attribution + message) by bumping a counter
Clears entries when an applet rebuilds successfully (old bundle no longer exists)
Entries include timestamp, source, applet attribution, message, optional stack, and occurrence count

CLI commands (server/cli.ts)

moi logs [--json] [--clear] — display or clear the error journal
moi call <module>/<fn> '[args]' — invoke a server function with JSON arguments, returns devalue-encoded result
moi shot widget|view <name> — screenshot an applet at its real grid size (or full panel for views)

Browser-side error reporting (client/lib/applet-log.ts)

Catches module load failures, render crashes, and window-level errors
Attributes errors to applets by matching stack frames against bundle URLs
POSTs errors to POST /api/workspaces/:id/applet-log with per-error cooldown (5s) to prevent storms
Always-on reporting: errors are recorded even if the user doesn't immediately report them

Applet screenshot support (client/lib/applet-shot.tsx, server/applet-shot.ts)

Mounts applets offscreen in the workspace's themed wrapper (so fonts/colors apply correctly)
Waits for mount-time data fetches to settle (configurable, default 1.5s)
Rasterizes with modern-screenshot and measures content overflow
Returns PNG + render facts (box dimensions, content size) to the CLI
Uses relay pattern (same as Scratchpad) to broadcast requests to live tabs

Control port integration (server/control.ts)

Refactored sendScratch → generic sendControl to handle multiple request types
Added handlers for logs, call, and shot request types
moi bundle output now nudges toward moi logs when errors are on record

Type definitions (lib/types.ts)

AppletKind — 'widget' | 'view'
AppletLogSource — 'build' | 'load' | 'render' | 'window' | 'rpc'
AppletLogEntry — full journal entry shape with dedup count
ShotRequest / ShotResult — screenshot request/response types

Generic relay utility (server/relay.ts)

Extracted relay mechanics (pending map, timeout, first-reply-wins) into reusable createRelay<Res>
Used by both Scratchpad (scratchpad-relay.ts) and applet shots (applet-shot.ts)

Error boundary updates (client/components/WidgetErrorBoundary.tsx)

Now reports render crashes to the applet log
Accepts kind (widget/view) and workspaceId for attribution
Optional onError callback for callers like applet shots

RPC error recording (server/api.ts)

Server-function failures now recorded in the journal with module/function attribution
Allows moi call to surface issues before the browser sees them

https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

The agent that builds applets could not tell when they broke: a widget compiles, then fails to load in the browser, crashes on render, or its server function throws — and nothing reaches the agent until the user complains. Close the loop with two legs (spec: docs/self-correction.md): - moi debug logs — an applet error journal under the new experimental `moi debug` command group. A per-workspace, in-memory ring buffer fed server-side (RPC failures, build failures) and by the browser (load failures, render crashes via the error boundary, window errors attributed by bundle-URL stack match, POSTed to /applet-log with a hostile-input guard). Identical errors dedup into a count; a successful rebuild clears the applet's standing entries; `moi bundle` ends with a nudge when errors remain on record. - moi call-server-fn — invoke one .server.ts function directly. Each invocation runs in an EPHEMERAL one-shot worker: spawn, single call, kill — fully isolated from the warm pool the widgets use, with the same env injection, module loading, timeout, and devalue wire format. Args as one JSON array; results render via Bun.inspect with duration. The moi-workspace skill grows a "Verify your work" section teaching the loop (bump to 0.6.0), with `moi debug` marked experimental. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

molefrog · 2026-07-18T17:25:03Z

 - `moi refresh` — re-fetch widget/view data without rebuilding (use after you mutated data the
  widgets read — DB rows, files, external API records — so the displayed values catch up)
+- `moi call-server-fn <module>/<fn> '[args]'` — invoke a `.server.ts` function directly
+  (smoke test; see Verify your work)


remove "see..."

molefrog · 2026-07-18T17:25:13Z

  widgets read — DB rows, files, external API records — so the displayed values catch up)
+- `moi call-server-fn <module>/<fn> '[args]'` — invoke a `.server.ts` function directly
+  (smoke test; see Verify your work)
+- `moi debug logs` — applet runtime errors on record (experimental; see Verify your work)


Suggested change

- `moi debug logs` — applet runtime errors on record (experimental; see Verify your work)

- `moi debug logs` — applet runtime errors on record (experimental)

Address PR #30 review: drop the "see Verify your work" cross-references from the CLI list, and rewrite the section (now "Debugging applets") so call-server-fn and debug logs are offered as feedback channels the agent can reach for, rather than a mandated verify-every-time checklist. The ambient signals (bundle footer nudge, always-on browser reporting) carry the discovery instead. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

Two upgrades to the window-error half of the applet journal: - Attribute 'error' events via ErrorEvent.filename first — a structured URL with no per-browser stack-format quirks, which also covers throws that carry no usable stack (`throw 'string'`). The stack-text scan stays as the fallback (handler threw inside vendor code) and as the only path for unhandled rejections, which have no filename. - Close the flood hole: the per-error cooldown keys on the message, so an error whose text varies each occurrence (timestamps, ids) bypassed it — and the server dedup — turning a tight throw loop into a POST storm. A global cap (30 reports/min, silent drops) bounds the total regardless of message shape. The URL matcher is extracted and unit-tested against bare filenames, Chrome/Firefox stack shapes, and near-miss URLs (vendor, rpc). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

Address PR #30 review: drop the "see Verify your work" cross-references from the CLI list, and rewrite the section (now "Debugging applets") so call-server-fn and debug logs are offered as feedback channels the agent can reach for, rather than a mandated verify-every-time checklist. The ambient signals (bundle footer nudge, always-on browser reporting) carry the discovery instead. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

molefrog force-pushed the claude/agent-self-correction-aegbzt branch from 7f037df to 174ab37 Compare July 17, 2026 09:12

molefrog commented Jul 18, 2026

View reviewed changes

claude added 3 commits July 18, 2026 17:29

Drop the timeout-duration note from the skill's call-server-fn entry

03d79d9

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M8rSCv39HrkfaH23tceb2q

molefrog merged commit 05cc056 into main Jul 18, 2026
1 check passed

molefrog deleted the claude/agent-self-correction-aegbzt branch July 18, 2026 19:01

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add self-correction loop: logs, call, and shot commands - #30

Add self-correction loop: logs, call, and shot commands#30
molefrog merged 4 commits into
mainfrom
claude/agent-self-correction-aegbzt

molefrog commented Jul 7, 2026

Uh oh!

molefrog Jul 18, 2026

Uh oh!

molefrog Jul 18, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

	- `moi debug logs` — applet runtime errors on record (experimental; see Verify your work)
	- `moi debug logs` — applet runtime errors on record (experimental)

Conversation

molefrog commented Jul 7, 2026

Summary

Key Changes

Uh oh!

molefrog Jul 18, 2026

Choose a reason for hiding this comment

Uh oh!

molefrog Jul 18, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants