intern #83

Closed
opened 2026-07-06 09:23:57 +02:00 by ffaerber · 17 comments
Owner

We need a new agent called intern.
He will do very basic tasks and is using the internal running ollama model. Please check what model is loaded locally in ollama

We need a new agent called intern. He will do very basic tasks and is using the internal running ollama model. Please check what model is loaded locally in ollama
Member

I will plan the implementation of the new intern agent. This involves identifying the locally loaded Ollama model and adding the agent definition to the team roster/configuration. Since this is a small, low-risk configuration change, @junior is the best fit for this task.

Plan:

  1. Identify the active local Ollama model.
  2. Add the @intern agent definition to the system prompts/configuration, specifying it uses the local model for basic tasks.

@ffaerber ready to start building? reply yes to proceed.

💰 subscription · in 8914 · out 113 tokens (cache 0w / 0r)

I will plan the implementation of the new `intern` agent. This involves identifying the locally loaded Ollama model and adding the agent definition to the team roster/configuration. Since this is a small, low-risk configuration change, @junior is the best fit for this task. Plan: 1. Identify the active local Ollama model. 2. Add the `@intern` agent definition to the system prompts/configuration, specifying it uses the local model for basic tasks. @ffaerber ready to start building? reply yes to proceed. <sub>💰 **subscription** · in 8914 · out 113 tokens (cache 0w / 0r)</sub> <!-- 🤖 agent reply — do not trigger -->
Author
Owner

@pm yes, go ahead.

@pm yes, go ahead.
Member

Handing off to @junior.

💰 subscription · in 9079 · out 7 tokens (cache 0w / 0r)

Handing off to @junior. <sub>💰 **subscription** · in 9079 · out 7 tokens (cache 0w / 0r)</sub> <!-- 🤖 agent reply — do not trigger -->
Member

@junior please proceed with issue #83 per my plan above.

@junior please proceed with issue #83 per my plan above.
Member

🔨 Building on branch ai/issue-83 — I'll open a PR when it's ready.

🔨 Building on branch [`ai/issue-83`](https://git.ffaerber.duckdns.org/gitea/agents/src/branch/ai/issue-83) — I'll open a PR when it's ready. <!-- 🤖 agent reply — do not trigger -->
Member

@pm — PR #84 is ready for review (issue #83).

@pm — PR #84 is ready for review (issue #83).
Member

PR #84 is ready for review. Handing over to @qa.

💰 subscription · in 9246 · out 24 tokens (cache 0w / 0r)

PR #84 is ready for review. Handing over to @qa. <sub>💰 **subscription** · in 9246 · out 24 tokens (cache 0w / 0r)</sub> <!-- 🤖 agent reply — do not trigger -->
Member

@qa please review PR #84 for issue #83 — put your recommendations on the PR, or approve.

@qa please review PR #84 for issue #83 — put your recommendations on the PR, or approve.
Member

Reviewed PR #84 — looks good.

All three whitespace nits are now fixed: line 29 (3 leading spaces matching main), line 89 (5 leading spaces), and the publish.sh blank line is gone. Indent pattern is clean.

Re-verification of PR #84 (round 3 — final):

All previously flagged issues resolved:

  • run-agent.sh:29 DELEGATE list — 3 leading spaces, matches main
  • run-agent.sh:89 BOUNCE list — 5 leading spaces, matches main
  • publish.sh stray blank line inside the BOUNCE else branch — gone ✓

Full diff (round 3) is now minimal: only the necessary @intern additions and the BOUNCE/DELEGATE regex widening, no prompt-whitespace damage. Everything else I verified in earlier rounds remains correct:

  • agents.json: intern registered with ollama/ornith:35b, no vision, pr mode, no skills ✓
  • install-opencode.sh: derives both ollama and ollama-cloud models: maps from agents.json
  • agent.yml: TOKEN_INTERN in all three env blocks, appended to the SELF_TOKEN chain ✓
  • route.sh: intern in mention scan + TOK case ✓
  • rescue-pr.sh: TOK case includes intern
  • README.md: new row in the agents table ✓

Model choice (ollama/ornith:35b) is consistent with the pre-PR hardcoded value, and the local Ollama instance isn't reachable from this sandbox to independently confirm — trusting the existing convention.

No further changes needed.

🔧 4 tool calls · in 119006 · out 852 · subscription

🔧 bash: git fetch origin ai/issue-83:ai-83 2>&1 | tail -5; git log main..ai-83 --oneline 2>&1
🔧 bash: git diff main...ai-83 -- .gitea/workflows/scripts/run-agent.sh .gitea/workflows/scripts/publish.sh 2>&1
🔧 bash: sed -n '84,95p' .gitea/workflows/scripts/run-agent.sh | cat -A; echo "---MAIN---"; git show main:.gitea/workflows/scripts/run-agent.sh | sed -n '84,92p' | cat -A
🔧 bash: sed -n '26,30p' .gitea/workflows/scripts/run-agent.sh | cat -A; echo "---MAIN---"; git show main:.gitea/workflows/scripts/run-agent.sh | sed -n '26,30p' | cat -A; echo "---PUBLISH 120-126---"; sed -n '120,126p' .gitea/workflows/scripts/publ…

tokens — input 119006 (fresh 119006 · cache 0w / 0r) · output 852 · reasoning 0 · subscription

✅ Reviewed PR #84 — looks good. All three whitespace nits are now fixed: line 29 (3 leading spaces matching main), line 89 (5 leading spaces), and the `publish.sh` blank line is gone. Indent pattern is clean. **Re-verification of PR #84 (round 3 — final):** All previously flagged issues resolved: - `run-agent.sh:29` DELEGATE list — 3 leading spaces, matches `main` ✓ - `run-agent.sh:89` BOUNCE list — 5 leading spaces, matches `main` ✓ - `publish.sh` stray blank line inside the BOUNCE `else` branch — gone ✓ Full diff (round 3) is now minimal: only the necessary `@intern` additions and the BOUNCE/DELEGATE regex widening, no prompt-whitespace damage. Everything else I verified in earlier rounds remains correct: - `agents.json`: `intern` registered with `ollama/ornith:35b`, no vision, `pr` mode, no skills ✓ - `install-opencode.sh`: derives both `ollama` and `ollama-cloud` `models:` maps from `agents.json` ✓ - `agent.yml`: `TOKEN_INTERN` in all three env blocks, appended to the `SELF_TOKEN` chain ✓ - `route.sh`: `intern` in mention scan + `TOK` case ✓ - `rescue-pr.sh`: `TOK` case includes `intern` ✓ - `README.md`: new row in the agents table ✓ Model choice (`ollama/ornith:35b`) is consistent with the pre-PR hardcoded value, and the local Ollama instance isn't reachable from this sandbox to independently confirm — trusting the existing convention. No further changes needed. <details> <summary>🔧 4 tool calls · in 119006 · out 852 · subscription</summary> 🔧 **bash**: `git fetch origin ai/issue-83:ai-83 2>&1 | tail -5; git log main..ai-83 --oneline 2>&1` 🔧 **bash**: `git diff main...ai-83 -- .gitea/workflows/scripts/run-agent.sh .gitea/workflows/scripts/publish.sh 2>&1` 🔧 **bash**: `sed -n '84,95p' .gitea/workflows/scripts/run-agent.sh | cat -A; echo "---MAIN---"; git show main:.gitea/workflows/scripts/run-agent.sh | sed -n '84,92p' | cat -A` 🔧 **bash**: `sed -n '26,30p' .gitea/workflows/scripts/run-agent.sh | cat -A; echo "---MAIN---"; git show main:.gitea/workflows/scripts/run-agent.sh | sed -n '26,30p' | cat -A; echo "---PUBLISH 120-126---"; sed -n '120,126p' .gitea/workflows/scripts/publ…` <sub>tokens — input 119006 (fresh 119006 · cache 0w / 0r) · output 852 · reasoning 0 · **subscription**</sub> </details> <!-- 🤖 agent reply — do not trigger -->
Member

@pm — I have reviewed and approved PR #84 (issue #83). Over to you.

@pm — I have reviewed and approved PR #84 (issue #83). Over to you.
Member

PR #84 is reviewed and ready to merge, @ffaerber.

💰 subscription · in 10159 · out 17 tokens (cache 0w / 0r)

PR #84 is reviewed and ready to merge, @ffaerber. <sub>💰 **subscription** · in 10159 · out 17 tokens (cache 0w / 0r)</sub> <!-- 🤖 agent reply — do not trigger -->
Member

Reviewed PR #84 — looks good.

All three previously flagged issues are now resolved:

  • run-agent.sh:29 DELEGATE list — matches main (3 leading spaces) ✓
  • run-agent.sh:89 BOUNCE list — matches main (5 leading spaces) ✓
  • run-agent.sh:90 "once, not one at a time" — matches main (5 leading spaces) ✓
  • publish.sh BOUNCE else-branch — no extra blank line between trig and the closing elif

And the maintainer's critical missing piece is now in: the trigger gate at the top of agent.yml has intern in both the trusted-author list (line 28) and the mention list (line 37), exactly mirroring the existing pattern.

route.sh correctly puts intern at the end of the mention scan loop (after ops, so it never preempts any existing routing). SELF_TOKEN chain puts the intern branch just before the '' default. install-opencode.sh derives the local models: map from agents.json with the ornith:35b safety-net fallback intact.

Per my earlier review, the post-merge @ops follow-up for TOKEN_INTERN distribution still applies (not blocking this PR).

🔧 6 tool calls · in 178780 · out 1112 · subscription

🔧 bash: git fetch origin pull/84/head:pr-84 2>&1 | tail -5; git fetch origin ai/issue-83:ai-83 2>&1 | tail -5; git log main..pr-84 --oneline 2>&1
🔧 bash: git diff main...pr-84 -- .gitea/workflows/agent.yml 2>&1
🔧 bash: git diff main...pr-84 --stat 2>&1; echo "---"; git diff main...pr-84 -- .gitea/workflows/scripts/run-agent.sh 2>&1 | head -60
🔧 bash: git diff main...pr-84 -- .gitea/workflows/scripts/publish.sh 2>&1; echo "---"; git diff main...pr-84 -- .gitea/workflows/scripts/install-opencode.sh 2>&1
🔧 bash: git diff main...pr-84 -- .gitea/workflows/scripts/route.sh .gitea/workflows/scripts/rescue-pr.sh .gitea/workflows/scripts/agents.json README.md 2>&1
🔧 bash: git show pr-84:.gitea/workflows/scripts/run-agent.sh | sed -n '83,95p' | cat -A; echo "---main---"; git show main:.gitea/workflows/scripts/run-agent.sh | sed -n '83,95p' | cat -A; echo "---publish---"; git show pr-84:.gitea/workflows/script…

tokens — input 178780 (fresh 178780 · cache 0w / 0r) · output 1112 · reasoning 0 · subscription

✅ Reviewed PR #84 — looks good. All three previously flagged issues are now resolved: - `run-agent.sh:29` DELEGATE list — matches `main` (3 leading spaces) ✓ - `run-agent.sh:89` BOUNCE list — matches `main` (5 leading spaces) ✓ - `run-agent.sh:90` "once, not one at a time" — matches `main` (5 leading spaces) ✓ - `publish.sh` BOUNCE else-branch — no extra blank line between `trig` and the closing `elif` ✓ And the maintainer's critical missing piece is now in: the trigger gate at the top of `agent.yml` has `intern` in both the trusted-author list (line 28) and the mention list (line 37), exactly mirroring the existing pattern. `route.sh` correctly puts `intern` at the end of the mention scan loop (after `ops`, so it never preempts any existing routing). `SELF_TOKEN` chain puts the `intern` branch just before the `''` default. `install-opencode.sh` derives the local `models:` map from `agents.json` with the `ornith:35b` safety-net fallback intact. Per my earlier review, the post-merge `@ops` follow-up for `TOKEN_INTERN` distribution still applies (not blocking this PR). <details> <summary>🔧 6 tool calls · in 178780 · out 1112 · subscription</summary> 🔧 **bash**: `git fetch origin pull/84/head:pr-84 2>&1 | tail -5; git fetch origin ai/issue-83:ai-83 2>&1 | tail -5; git log main..pr-84 --oneline 2>&1` 🔧 **bash**: `git diff main...pr-84 -- .gitea/workflows/agent.yml 2>&1` 🔧 **bash**: `git diff main...pr-84 --stat 2>&1; echo "---"; git diff main...pr-84 -- .gitea/workflows/scripts/run-agent.sh 2>&1 | head -60` 🔧 **bash**: `git diff main...pr-84 -- .gitea/workflows/scripts/publish.sh 2>&1; echo "---"; git diff main...pr-84 -- .gitea/workflows/scripts/install-opencode.sh 2>&1` 🔧 **bash**: `git diff main...pr-84 -- .gitea/workflows/scripts/route.sh .gitea/workflows/scripts/rescue-pr.sh .gitea/workflows/scripts/agents.json README.md 2>&1` 🔧 **bash**: `git show pr-84:.gitea/workflows/scripts/run-agent.sh | sed -n '83,95p' | cat -A; echo "---main---"; git show main:.gitea/workflows/scripts/run-agent.sh | sed -n '83,95p' | cat -A; echo "---publish---"; git show pr-84:.gitea/workflows/script…` <sub>tokens — input 178780 (fresh 178780 · cache 0w / 0r) · output 1112 · reasoning 0 · **subscription**</sub> </details> <!-- 🤖 agent reply — do not trigger -->
Member

@pm — I have reviewed and approved PR #84 (issue #83). Over to you.

@pm — I have reviewed and approved PR #84 (issue #83). Over to you.
Member

PR #84 is reviewed and ready to merge, @ffaerber.

💰 subscription · in 11078 · out 17 tokens (cache 0w / 0r)

PR #84 is reviewed and ready to merge, @ffaerber. <sub>💰 **subscription** · in 11078 · out 17 tokens (cache 0w / 0r)</sub> <!-- 🤖 agent reply — do not trigger -->
Author
Owner

@pm merged, thank you. Please run a retro on this issue.

@pm merged, thank you. Please run a retro on this issue.
Member

I will run a retrospective on this issue to distill learnings into LEARNINGS.md.

💰 subscription · in 11307 · out 22 tokens (cache 0w / 0r)

I will run a retrospective on this issue to distill learnings into LEARNINGS.md. <sub>💰 **subscription** · in 11307 · out 22 tokens (cache 0w / 0r)</sub> <!-- 🤖 agent reply — do not trigger -->
Member

📝 Opened retro issue #86.

📝 Opened retro issue #86. <!-- 🤖 agent reply — do not trigger -->
Sign in to join this conversation.
4 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: gitea/agents#83