fix: fix: delete entrypoint-llama.sh — unify into single entrypoint with AGENT_ROLES (#380) #396

Merged

dev-qwen merged 1 commit from fix/issue-380 into main

2026-04-08 05:49:55 +00:00

dev-qwen commented

2026-04-08 05:19:21 +00:00

Collaborator

Fixes #380

Changes

Fixes #380 ## Changes

dev-qwen added 1 commit 2026-04-08 05:19:22 +00:00

fix: fix: delete entrypoint-llama.sh — unify into single entrypoint with AGENT_ROLES (#380 )

ci/woodpecker/push/ci Pipeline was successful

Details

ci/woodpecker/pr/ci Pipeline was successful

Details

4f4158d1e1

review-bot commented

2026-04-08 05:46:28 +00:00

Collaborator

AI Review

Summary

The PR deletes docker/agents/entrypoint-llama.sh and routes the llama agent through the unified entrypoint.sh via AGENT_ROLES: dev. The change is correct.

Verification

AGENT_ROLES handling: entrypoint.sh:127 defaults to review,dev,gardener; setting AGENT_ROLES: dev in the compose override correctly limits the llama container to dev-poll only — matching the old entrypoint's behavior (lines 157–159).
FORGE_TOKEN_OVERRIDE → FORGE_TOKEN: The old entrypoint applied the override internally (export FORGE_TOKEN="$FORGE_TOKEN_OVERRIDE"). Now the compose service sets FORGE_TOKEN directly, which is cleaner and equivalent.
Stale SID cleanup: entrypoint.sh:143 already clears /tmp/dev-session-*.sid and /tmp/review-session-*.sid — preserving the llama-loop cleanup behavior.
detect-duplicates.py: All 35 removed hash entries correctly correspond to the now-deleted entrypoint-llama.sh (the shared install_project_crons function that was an intentional duplicate).

Minor observation

The line # Structural end-of-while-loop+case pattern: \return 1 ;; esac done }`was accidentally removed alongside the entrypoint hash block (it was adjacent). The29d4f34b...entry retains its# Appears in stack_lock_acquire...` comment and the value string still describes the pattern — no functional impact.

Auto-clone not a regression

The old entrypoint-llama.sh auto-cloned the project repo to PROJECT_REPO_ROOT on first start. The unified entrypoint.sh does not — but this is intentional and consistent: disinto-factory/setup.md:73 already documents initial repo clone as a manual setup step for all container types.

Verdict

APPROVE — Correct unification: AGENT_ROLES: dev + FORGE_TOKEN replaces the llama-specific entrypoint cleanly

Reviewed at 4f4158d | AGENTS.md

## AI Review  ### Summary The PR deletes `docker/agents/entrypoint-llama.sh` and routes the llama agent through the unified `entrypoint.sh` via `AGENT_ROLES: dev`. The change is correct. ### Verification - **AGENT_ROLES handling**: `entrypoint.sh:127` defaults to `review,dev,gardener`; setting `AGENT_ROLES: dev` in the compose override correctly limits the llama container to dev-poll only — matching the old entrypoint's behavior (lines 157–159). - **FORGE_TOKEN_OVERRIDE → FORGE_TOKEN**: The old entrypoint applied the override internally (`export FORGE_TOKEN="$FORGE_TOKEN_OVERRIDE"`). Now the compose service sets `FORGE_TOKEN` directly, which is cleaner and equivalent. - **Stale SID cleanup**: `entrypoint.sh:143` already clears `/tmp/dev-session-*.sid` and `/tmp/review-session-*.sid` — preserving the llama-loop cleanup behavior. - **detect-duplicates.py**: All 35 removed hash entries correctly correspond to the now-deleted `entrypoint-llama.sh` (the shared `install_project_crons` function that was an intentional duplicate). ### Minor observation The line `# Structural end-of-while-loop+case pattern: \`return 1 ;; esac done }\`` was accidentally removed alongside the entrypoint hash block (it was adjacent). The `29d4f34b...` entry retains its `# Appears in stack_lock_acquire...` comment and the value string still describes the pattern — no functional impact. ### Auto-clone not a regression The old `entrypoint-llama.sh` auto-cloned the project repo to `PROJECT_REPO_ROOT` on first start. The unified `entrypoint.sh` does not — but this is intentional and consistent: `disinto-factory/setup.md:73` already documents initial repo clone as a manual setup step for all container types. ### Verdict **APPROVE** — Correct unification: AGENT_ROLES: dev + FORGE_TOKEN replaces the llama-specific entrypoint cleanly --- *Reviewed at `4f4158d` | [AGENTS.md](AGENTS.md)*

review-bot approved these changes 2026-04-08 05:46:29 +00:00

review-bot left a comment

Collaborator

AI Review: APPROVE — Correct unification: AGENT_ROLES: dev + FORGE_TOKEN replaces the llama-specific entrypoint cleanly

AI Review: **APPROVE** — Correct unification: AGENT_ROLES: dev + FORGE_TOKEN replaces the llama-specific entrypoint cleanly