xonecas.io
64875
Mc

mini-coder / commit c99c5ea

chore: update agents and todo notes

xonecas2026-03-11 c99c5ea440b323c9a801978c95ec5a2c7a6f3df7 parent c1d12fb

5 files changed

.agents/agents/orc.md+1−1
@@ -3,7 +3,7 @@ description: Orchestrator — decomposes complex tasks and coordinates a fleet o
33 mode: primary
44 ---
55
6-You are **Orc**, an expert orchestration agent. Your superpower is breaking complex problems into well-scoped subtasks and delegating each to the right specialist subagent via the `subagent` tool.
6+You are in **Orchestration** mode, behave like an expert orchestration agent. Your superpower is breaking complex problems into well-scoped subtasks and delegating each to the right specialist subagent via the `subagent` tool.
77
88 ## Core principles
99
.agents/agents/writer.md+1−3
@@ -1,7 +1,6 @@
11 ---
22 description: Write short prose in a personal, informal style from topics or resources
33 mode: subagent
4-model: zen/claude-sonnet-4-6
54 ---
65
76 # Blog Writer Agent
@@ -41,5 +40,4 @@ You write short, personal, informal blog posts. Your output is a markdown draft
4140 - Markdown only.
4241 - Start with an H1 title.
4342 - Use headings and short paragraphs.
44-- End with a `Sources` section that lists URLs or references used.
45-- In a markdown file at repo root: `<title>.draft.md`
43+- In a markdown file at repo root: `<title>.draft.md`
.agents/commands/release.md+4−5
@@ -2,23 +2,23 @@
22 description: Create a new tagged npm release (patch/minor/major)
33 ---
44
5-
65 Create a new npm release for this repo. The bump type is: $1 (default to "patch" if not provided).
76
87 Steps to follow exactly:
8+
99 1. Normalize bump type to one of `patch|minor|major`; if empty, use `patch`.
1010 2. Verify git state before doing anything:
1111 - current branch must be `main`
1212 - working tree must be clean
1313 - run `git fetch origin --tags` and ensure `HEAD` equals `origin/main`
14- If any check fails, stop and report.
15-3. Run `bun run typecheck && bun run format && bun run lint && bun test` — stop and report failures.
14+ If any check fails, stop and report.
15+3. Run the full repo check suit — stop and report failures.
1616 4. Read `package.json` and capture both `name` and current `version`.
1717 5. Compute the next semver version by bumping the selected part.
1818 6. Guardrails before mutating files:
1919 - ensure git tag `v<new-version>` does not exist locally or on origin
2020 - ensure npm version does not already exist: `npm view <package-name>@<new-version> version`
21- If either exists, stop and report.
21+ If either exists, stop and report.
2222 7. Update the `version` field in `package.json` to `<new-version>`.
2323 8. Run `bun run build`.
2424 9. Commit all staged and unstaged changes with message: `chore: release v<new-version>`.
@@ -29,4 +29,3 @@ Steps to follow exactly:
2929 12. Post-push verification: run `npm view <package-name>@<new-version> version`.
3030 - If it already exists, report that publish already happened and do **not** ask user to publish again.
3131 - If it does not exist, ask user to run `npm publish` (passkey auth step).
32-
.agents/commands/search.md+0−15
@@ -1,15 +0,0 @@
1----
2-description: Search the codebase for a topic using a subagent
3-model: zen/claude-haiku-4-5
4-context: fork
5----
6-
7-
8-Search the codebase for: $ARGUMENTS
9-
10-Use glob, grep, and read tools to explore thoroughly. Report:
11-1. All relevant files and locations found
12-2. Key code snippets with file paths and line numbers
13-3. A short summary of what you found
14-
15-Be exhaustive but concise. No edits — read only.
TODO.md+49−0
@@ -1,5 +1,54 @@
11 # TODO
22
3+## Custom subagents are broken:
4+
5+They are being called with the wrong arguments, or something is going wrong in the call, see the errors bellow:
6+
7+Claude attempt:
8+
9+```
10+◆ Now I have all the context I need. Handing off to the writer! 🖊️
11+ ⇢ [toolu_] — Write an exciting blog post draft announcing that mini-co…
12+⣟ subagenterror: Script not found "zen/claude-sonnet-4-6"
13+ ✖ Subagent subprocess produced no output (exit code 1)
14+```
15+
16+Kimi 2.5 attempts:
17+
18+```
19+ I have a good understanding of the project now. Let me delegate to the writer agent to create an exciting blog post about reaching v1 and stable release. I should provide the writer with context about what features we've implemented.
20+ ⇢ [functi] — Write an exciting blog post draft announcing that mini-co…
21+◆ error: File not found "zen/kimi-k2.5"
22+ ✖ Subagent subprocess produced no output (exit code 1)
23+
24+ The subagent didn't produce output. Let me try again with a more direct prompt and see if there's an issue with the writer agent.
25+◆ Let me check the writer agent configuration and try again:
26+ ← read .agents/agents/writer.md
27+ · .agents/agents/writer.md 44 lines
28+◆
29+ The writer agent looks fine. The issue might be with how the subagent is being invoked or there's some error. Let me try running the writer subagent again with a simpler prompt.
30+◆
31+ ⇢ [functi] — Write an exciting blog post draft about mini-coder reachi…
32+◆ error: File not found "zen/kimi-k2.5"
33+ ✖ Subagent subprocess produced no output (exit code 1)
34+
35+ The subagent is failing silently. Let me try to debug this by running the subagent command manually to see what's happening.
36+◆ Let me debug the subagent issue:
37+ $ cd ~/src/mini-coder && mc @writer "Write a 100 word blog post about c…
38+ timeout
39+ The user is asking me to delegate to the "writer" subagent to write a 100-word blog post about coding agents. I need to use the `subagent` tool with the `agentName` set to "writer" and provide an appropriate prompt.
40+ │ ⇢ [functi] — Write a 100 word blog post about coding agents. Informal,…
41+◆ error: File not found "zen/kimi-k2.5"
42+ │ ✖ Subagent subprocess produced no output (exit code 1)
43+```
44+
45+```
46+◆ Now let me try using the subagent correctly:
47+ ⇢ [functi] — Write an exciting blog post draft about mini-coder reachi…
48+◆ error: File not found "zen/kimi-k2.5"
49+ ✖ Subagent subprocess produced no output (exit code 1)
50+```
51+
352 ---
453
554 ## LSP Diagnostics (not very important, with tool-hooks and strong linting, this is not as necessary, but we should implement asap)