[liususan091219]

Summary

Pure variable rename. 60 insertions, 60 deletions. No behavior change, no logic touched — every line is just s/\bREPO\b/WORKSPACE/.

7 sites where a variable was named REPO/REPO_DIR/_REPO/repo_resolved but its value came from resolve_workspace() / resolveWorkspace() / WORKSPACE_DIR.resolve(). The misleading names made the v3 workspace-contract audit harder than it should have been — readers (humans and LLMs running cwd-resolution audits) had to trace each REPO back to its assignment site to discover it actually held workspace state. After this PR WORKSPACE means workspace, REPO_DIR means repo, and you can tell which from the name alone.

Renames

File	Old	New	Uses
src/agent-api.py:496-498	repo_resolved	workspace_resolved	3
src/send_allowlist.py:51+	_REPO	_WORKSPACE	5
src/slack-bridge.py:59+	REPO	WORKSPACE	12
src/dm-result.py:38, 187	REPO	WORKSPACE	2
src/discord-bridge.py:59+	REPO	WORKSPACE	16
src/telegram-bridge.py:37+	REPO	WORKSPACE	15
src/task-bridge.ts:18+	REPO_DIR	WORKSPACE_DIR	7

src/discord-bridge.py:L3685-3687 also has a comment referring to "REPO" by name when explaining why dm-result.py is resolved via __file__ — comment updated to read "WORKSPACE" so it stays in sync with the variable; the historical reference to PR #762 is preserved verbatim.

What is NOT renamed

• src/agent-api.py:93 REPO_DIR = Path(__file__).parent.parent — the name is accurate (it actually holds the repo path). Kept as-is.
• scripts/{presenter-mode,query-conversation,results-health}.sh — audit-flagged as cosmetic but these are part of PR [workspace-migration] fix(scripts): 5 sh scripts read/write workspace state, not repo (P1 batch) #1276 (5-scripts workspace-anchor fix), where the rename happens alongside the actual path fix. Bundling there keeps the test+script change together.
• 81 other "cosmetic" audit findings — different bug classes (LLM prose paths, test fixtures writing to real workspace, dual-read patterns). Not addressable by a simple rename; tracked separately in workspace/notes/cwd-audit-v3-2026-05-27.md.

Verification

$ git diff --stat
 src/agent-api.py       |  6 +++---
 src/discord-bridge.py  | 32 ++++++++++++++++----------------
 src/dm-result.py       |  4 ++--
 src/send_allowlist.py  | 10 +++++-----
 src/slack-bridge.py    | 24 ++++++++++++------------
 src/task-bridge.ts     | 14 +++++++-------
 src/telegram-bridge.py | 30 +++++++++++++++---------------
 7 files changed, 60 insertions(+), 60 deletions(-)

$ for f in src/agent-api.py src/send_allowlist.py src/slack-bridge.py src/dm-result.py src/discord-bridge.py src/telegram-bridge.py src/task-bridge.ts; do
    cnt=$(grep -cE '\b(REPO|REPO_DIR|_REPO|repo_resolved)\b' "$f")
    echo "$f: $cnt stray REPO-ish bare uses"
  done
src/agent-api.py: 4 remaining REPO-ish bare uses  ← all legitimate (real REPO_DIR at L93+, used at L129, L440)
src/send_allowlist.py: 0
src/slack-bridge.py: 0
src/dm-result.py: 0
src/discord-bridge.py: 0
src/telegram-bridge.py: 0
src/task-bridge.ts: 0

npx tsc --noEmit clean.

Method

Word-boundary regex substitution: re.sub(r'\b{OLD}\b', '{NEW}', src). This ensures docstring/comment text like "the repo" / "repo_root" / "REPO_DIR-elsewhere" (legitimate) wasn't touched. Each file was eyeballed before substitution to confirm no other identifier collides with the rename token.

Refs v3 workspace-contract audit, P3 cosmetic batch (88 sites flagged in audit; 7 are simple variable-name-lies addressable by rename — this PR. The rest fall under PR #1276 or remain in the audit backlog as their own bug classes).

🤖 Generated with Claude Code

[chetanunadkat]

Reviewed the rename. The 4 files agent-api.py, send_allowlist.py, slack-bridge.py, task-bridge.ts are clean — every renamed REPO/_REPO is resolve_workspace()-derived and all paths (tasks/, results/, state/, etc.) are genuine workspace state. Faithful rename, no behavior change. ✅

⚠️ Collision with #1274 on 3 lines — flagging before merge. This PR renames REPO → WORKSPACE at the same three .env-read sites that #1274 ("read .env from source tree, not workspace") is fixing as a bug:

• discord-bridge.py — env_file = REPO / ".env" → here becomes WORKSPACE / ".env"
• dm-result.py — REPO / ".env"
• telegram-bridge.py — load_dotenv(REPO / ".env")

.env lives in the source-tree root, not the workspace, so WORKSPACE / ".env" is the wrong target (#1274's point: these silently fail to load DISCORD_BOT_TOKEN / TEAM_TIER_OWNER / the Telegram token). The two PRs touch these exact lines with opposite intent:

• #1274 fixes the path (workspace → source tree), which makes the REPO name honest.
• #1333 renames the variable (REPO → WORKSPACE), cementing the workspace path.

They can't both land cleanly:

• #1274 first → this PR then renames a now-source-tree REPO to WORKSPACE (reverse name-lie + re-breaks .env loading).
• #1333 first → #1274 must rebase onto the new WORKSPACE name.

Suggest: drop the 3 .env sites from this rename (leave them to #1274, which makes REPO honest by pointing it at the source tree), or land #1274 first and rebase. The other 4 files are unaffected either way.

(Cold review from the Sutando proactive loop — context: verified against current main + #1274's diff.)

[amkunte]

LGTM — pure rename, +60/-60, zero behavior change. The name-truth audit is exactly the kind of cleanup that's easy to put off until a future debugger spends an hour tracing REPO_DIR back to a resolve_workspace() call site.

Strong points:

• The table in the PR body listing every site's old/new/uses field makes the change auditable in one read. Pure mechanical s/REPO/WORKSPACE/ in places where the value was already a workspace path.
• All 7 files cleanly bucket "value comes from resolve_workspace()" → no false-positives renaming an actual repo path.
• Doesn't touch REPO_DIR-named-things that actually hold repo paths (e.g. install-claude-hooks.sh:54's REPO_DIR). Discipline.

One thing to verify post-merge (non-blocking):
If anything outside this PR's diff was grep-importing REPO symbols from these files, the rename would break it. Diff is all module-local variables (not from X import REPO), so very unlikely — but worth a grep -r "from agent_api import REPO\|from discord_bridge import REPO\|..." across the repo as a 10-second sanity check before merge. Nothing in the immediate diff suggests cross-module use.

Excellent maintenance PR — costs nothing to land, makes every future audit cheaper. 👍

[sonichi]

LGTM — verified the "no behavior change" claim holds across all 7 files.

Since this renames the variable at its definition (REPO/_REPO = resolve_workspace() → WORKSPACE/_WORKSPACE), the only real risk is an unrenamed straggler reference (would become an undefined-name crash) or a name collision with a pre-existing WORKSPACE. Checked both on the branch head:

• Zero leftover bare REPO/_REPO tokens in any of the 7 files (excluding the unrelated SUTANDO_REPO_DIR/REPO_DIR/REPO_ROOT names) — rename is complete, no dangling refs.
• No pre-existing WORKSPACE/_WORKSPACE in any of the 7 files on main — no shadowing/collision introduced.
• Value is identical (resolve_workspace()), and the dm-fallback comment text in discord-bridge.py was updated in lockstep (REPO point at... → WORKSPACE point at...), so the comment no longer lies either.

Per-file convention is preserved (module-private _WORKSPACE in send_allowlist.py mirroring its original _REPO; plain WORKSPACE elsewhere). Pure cosmetic cleanup that removes the long-standing "this var says REPO but holds the workspace path" foot-gun. No blockers. (Reviewing as Sutando/Pro — no merge authority; flagging for Chi.)

[liususan091219]

Closing per Susan 2026-05-29 — Qingyun is redesigning the workspace migration approach. The 15-PR set will be re-applied (or rebuilt to match the new design) after that lands. Branch is preserved; can be reopened or cherry-picked when ready.