External/roboco
1
0
Fork 0
mirror of https://github.com/rennf93/roboco.git synced 2026-08-03 07:23:24 +02:00
Code Issues Packages Projects Releases Wiki Activity

feat(content): anti-soup guard on the flow verbs' free-text

Browse Source 
Extends structured-content enforcement from the content tools to every
flow verb that carries agent free-text, closing the last hole where a
dev/PM could pass word soup: i_am_blocked(reason), i_am_done(notes),
submit_up/submit_root/complete(notes), escalate_up/escalate_to_ceo(reason),
pass_review(notes), fail_review/pr_fail(issues), pr_pass(notes),
i_documented(notes), delegate(title/description). Plans (i_will_plan /
i_will_work_on) keep their existing >=150-char approach + sub_task gates
and are skipped here so recovery re-entry with thin values still works.

Shared helpers on the choreographer: _free_text_soup (bare envelope, list
aware) and _soup_or_decision_env (folds the soup check into a verb's
existing spec-gate return so no verb gains a return or tips the xenon
bound). reject_trivial now also catches all-filler multi-token strings.

base.md documents the broadened rule for agents.
This commit is contained in:
Renn F
2026-06-21 05:11:20 +02:00
parent 96ca53eff2
commit 4c85cc6dfa
13 changed files with 432 additions and 59 deletions
Download Patch File Download Diff File Expand all files Collapse all files
agents/prompts/base.md
+1 -1
View File
@@ -78,7 +78,7 @@ Channel arguments take the slug **without** the `#` prefix: `"backend-cell"`, no
- `env`/`printenv` is **denied** — secrets are not readable from your container.
- `Edit`/`Write` are scoped to your workspace: `/data/workspaces/{project}/{team}/{your-slug}/`.
- Subagents (the `Agent` tool, where granted) are for **parallel research only** — fanning out to read multiple files at once. They are NOT a way to delegate your actual task to another instance of yourself.
- **Everything you write is validated — no filler, no word soup, anywhere.** `say`/`dm`/`note`/`progress`/`notify` reject empty/placeholder text (`asdf`, `wip`, `tbd`, `...`); state what actually happened. Structured artifacts must fill their named fields, never a flat phrase: PR reviews take `findings` (each `{file, line, severity, expected, actual}`), QA takes `ac_verdicts` (one per criterion), `decision`/`reflect` take their structured fields, task drafts take `objective`/`the_work`/`acceptance_criteria`.
- **Everything you write is validated — no filler, no word soup, anywhere.** This holds for *every* free-text field, not just chat: content tools (`say`/`dm`/`note`/`progress`/`notify`/`pitch`/`pr_update`/`open_session`) AND the flow verbs' text — `i_am_blocked(reason=...)`, `i_am_done(notes=...)`, `submit_up`/`submit_root`/`complete(notes=...)`, `escalate_up`/`escalate_to_ceo(reason=...)`, `fail_review`/`pr_fail(issues=[...])`, `pass_review(notes=...)`, `delegate(title=, description=)`. Empty/placeholder text (`asdf`, `wip`, `tbd`, `n/a`, `...`, `x`) and all-filler strings (`wip wip`) are rejected with a remediable envelope; state what actually happened. Structured artifacts must fill their named fields, never a flat phrase: PR reviews take `findings` (each `{file, line, severity, expected, actual}`), QA takes `ac_verdicts` (one per criterion), `decision`/`reflect` take their structured fields, task drafts take `objective`/`the_work`/`acceptance_criteria`.
## Branch and commit conventions (handled by the gateway)