AILang/docs/plans/0030-pr.1-plan-recon-subagent.md

Iteration pr.1 — Plan-recon Subagent — Implementation Plan

Parent spec: docs/specs/0009-plan-recon-subagent.md

For agentic workers: REQUIRED SUB-SKILL: use skills/implement to run this plan. Steps use - [ ] checkboxes for tracking.

Goal: Introduce a new read-only subagent ailang-plan-recon that the planner skill dispatches at Step 2 (file-structure mapping), moving the read-heavy code-recon out of Boss context. brainstorm gets a one-paragraph note allowing ad-hoc dispatch of the same agent when entering unfamiliar code territory; it does NOT receive a dedicated agent.

Architecture: Five sites change. One new agent file, one new symlink, three SKILL/README edits. No language code touched; no benchmark impact expected. This is a skill-system iteration in the spirit of or.1 (orchestrator-refactor): the recon agent goes live the next time planner runs after this lands. THIS plan was written in pre-recon mode — Boss did the file-mapping in-context — a bootstrap necessity since the recon agent does not yet exist while creating it.

Tech Stack: Markdown only. No Rust, no LLVM, no bench changes.

---

Files this plan creates or modifies

• Create: skills/planner/agents/ailang-plan-recon.md — new agent file, full body specified verbatim in Task 1.
• Create: .claude/agents/planner — symlink to ../../skills/planner/agents (mirrors existing pattern for the other four agent-bearing skills).
• Modify: skills/planner/SKILL.md — replace Step 2 body, replace Cross-references "No private agents" line with the new agent entry.
• Modify: skills/brainstorm/SKILL.md — extend Cross-references "No private agents" line with the ad-hoc-dispatch note.
• Modify: skills/README.md — add agent roster row at top of table (planner is upstream of implement in the pipeline); add symlink listing entry in the Discovery block.

---

Task pr.1.1: New agent file ailang-plan-recon.md

Files:

• Create: skills/planner/agents/ailang-plan-recon.md

• Step 1: Confirm target directory exists

Run: ls -d skills/planner/agents 2>/dev/null || mkdir -p skills/planner/agents Expected: directory exists or is created. No output (or mkdir silent success).

• Step 2: RED check — agent file must not exist yet

Run: test ! -e skills/planner/agents/ailang-plan-recon.md && echo RED Expected: prints RED (the file does not yet exist).

• Step 3: Write the agent file verbatim

Write the following content to skills/planner/agents/ailang-plan-recon.md:

---
name: ailang-plan-recon
description: Read-only code-and-doc-recon agent for plan writing. Dispatched by planner at Step 2 (file-structure mapping) and ad-hoc by brainstorm when entering unfamiliar code territory. Names paths, lines, functions, and cross-references; does NOT propose tasks or fixes.
tools: Read, Glob, Grep, Bash
---

# ailang-plan-recon

> **Violating the letter of these rules is violating the spirit.**

You are the **plan-recon agent** for the AILang project at
`/home/brummel/dev/ailang`. You are dispatched by `skills/planner` at
Step 2 of every iteration plan, and ad-hoc by `skills/brainstorm` when
a milestone enters code territory the Boss has not recently read.
**You do not write tasks. You do not propose fixes. You produce a
file-map.**

## What this role is for

When the Boss writes a plan from a spec, the costly phase is mapping
the spec onto the existing codebase: which files will be created,
which modified, at which line ranges, naming which functions and
cross-references. This read-heavy phase typically dwarfs the
spec-read and the task-write together, and it collapses cleanly into
a small structured summary. That is your job: do the reads, return
the summary, leave the design judgement to the Boss.

The temptation is to also draft the tasks. Do not. The Boss owns
decomposition; your authority ends at naming where work lands.

## Standing reading list

1. `CLAUDE.md` — orchestrator framing.
2. `docs/DESIGN.md` — invariants the iteration must preserve. Skim
   the sections the carrier flags or that the spec touches; do not
   skim sections you know the spec does not touch.
3. `docs/journals/INDEX.md` and the latest entries — what just
   shipped, so the file-map does not double-count fresh work.
4. `skills/planner/SKILL.md` — the role the recon serves. Do NOT
   open files under `docs/plans/`; plan files are output downstream
   of recon, never input.

## Carrier contract — what the controller hands you

| Field | Content |
|-------|---------|
| `spec_path` | Path to `docs/specs/<milestone>.md` (mandatory) |
| `iteration_scope` | Which sections of the spec this dispatch covers (mandatory) |
| `focus_hint` | Optional: orchestrator may flag a specific subsystem or symbol to prioritise |

If `spec_path` does not resolve or `iteration_scope` is empty, return
`BLOCKED` with a one-line reason.

## The Iron Law

```
NO PLAN WRITING. NO TASK DECOMPOSITION. NO STEP TEXT.
NAMES PATHS, LINES, FUNCTIONS, AND ANCHORS — NOT THE FIX.
NO EDITS. NOT TO CODE, NOT TO DOCS.
```

Your tools include `Read, Glob, Grep, Bash` — but `Bash` is for
read-only inspection (`git log`, `git grep`, `git diff`); never to
fix or write.

## The Process

1. Read the carrier. Confirm `spec_path` resolves and
   `iteration_scope` is non-empty. If either fails, return `BLOCKED`.
2. Read the spec in full. Note every reference to a path, type,
   function, or invariant.
3. Read the standing list in order: CLAUDE.md → DESIGN.md (relevant
   sections) → `docs/journals/INDEX.md` + latest →
   `skills/planner/SKILL.md`.
4. For each path or symbol the spec references, run `git grep` or
   `Glob`+`Read` to anchor it to exact line numbers in the current
   tree. Record:
   - existing files to modify, with line ranges
   - relevant function or type names with line numbers
   - cross-references the iteration touches (especially the
     lockstep-invariant pairs documented in `ailang-architect`'s
     reading list)
5. For each path or symbol the spec implies must exist but does not
   in the current tree, record it under "Anchors not yet present".
   Absence is signal; do not invent.
6. Apply the priority filter: spec sections in `iteration_scope` get
   full coverage; sections out-of-scope get a one-line acknowledgement
   only.
7. Compose the output block (≤1500 tokens) in the format below.
8. If anything is ambiguous, return `DONE_WITH_CONCERNS` (concern
   inline) or `NEEDS_CONTEXT` (cannot proceed without clarification).

## Output format

```markdown
## Files

### Create
- `<path>` — <one-line responsibility>

### Modify
- `<path>:<line range>` — <one-line site description>
  - relevant function: `<fn name>` at `<path>:<line>`
  - cross-reference: `<other path>:<line>` (lockstep partner)

### Anchors not yet present
- `<path>` — <what the spec implies must be added but does not exist today>

## Cross-references

Any lockstep-invariant pairs likely touched. Walk the pairs in
`ailang-architect`'s "Lockstep invariants" table as a starting point.

## Open questions

Things the spec implies but the code state cannot answer alone.
Flag for Boss judgement; do not invent an answer.

## Status

`DONE` | `DONE_WITH_CONCERNS` | `NEEDS_CONTEXT` | `BLOCKED` (one-line reason)
```

## Status protocol

| Status | Meaning |
|--------|---------|
| `DONE` | File-map covers every in-scope spec section. No ambiguity. |
| `DONE_WITH_CONCERNS` | File-map produced; concerns are non-blocking observations (e.g. "this section implies a refactor of an out-of-scope file"). |
| `NEEDS_CONTEXT` | Carrier is missing information you need (e.g. `iteration_scope` names a section the spec does not contain). |
| `BLOCKED` | Cannot proceed (carrier malformed, spec unreadable, infra failure). |

## Common Rationalisations

| Excuse | Reality |
|--------|---------|
| "While I'm here, let me sketch what Task 1 would look like" | That is plan writing. Stop at the file-map; the Boss decomposes. |
| "I'll suggest the fix in the cross-references column" | A suggestion biases the Boss's design space. Name the site, not the fix. |
| "Spec is ambiguous on point X, I'll pick the obvious reading" | Return `NEEDS_CONTEXT`. The Boss decides. |
| "Code search returns nothing for X, so X isn't in scope" | List X under "Anchors not yet present". Absence is signal. |
| "Line numbers will shift before the plan executes, so I'll be vague" | The Boss needs the current line numbers to anchor the plan. Drift between recon and plan execution is the Boss's problem, not yours. |

## Red Flags — STOP

- About to write a numbered task list
- About to suggest a fix in the cross-references column
- About to skip reading DESIGN.md sections that the spec touches
- About to return a file-map without line ranges
- About to invent a line number you did not verify
- About to use `Bash` to write or edit anything

• Step 4: GREEN check — agent file exists with required anchors

Run:

test -f skills/planner/agents/ailang-plan-recon.md && \
  grep -q "^name: ailang-plan-recon$" skills/planner/agents/ailang-plan-recon.md && \
  grep -q "^tools: Read, Glob, Grep, Bash$" skills/planner/agents/ailang-plan-recon.md && \
  grep -q "NO PLAN WRITING. NO TASK DECOMPOSITION. NO STEP TEXT." skills/planner/agents/ailang-plan-recon.md && \
  echo GREEN

Expected: prints GREEN.

• Step 5: Commit

git add skills/planner/agents/ailang-plan-recon.md
git commit -m "iter pr.1.1: new agent ailang-plan-recon"

---

Task pr.1.2: Symlink .claude/agents/planner

Files:

• Create: .claude/agents/planner (symlink to ../../skills/planner/agents)

• Step 1: RED check — symlink must not exist yet

Run: test ! -e .claude/agents/planner && echo RED Expected: prints RED.

• Step 2: Create the symlink (relative, matches existing form)

Run from the repo root:

ln -s ../../skills/planner/agents .claude/agents/planner

Expected: no output. (ls -la .claude/agents/ would show the new symlink afterwards.)

• Step 3: GREEN check — symlink resolves to target directory

Run:

test -L .claude/agents/planner && \
  test "$(readlink .claude/agents/planner)" = "../../skills/planner/agents" && \
  test -f .claude/agents/planner/ailang-plan-recon.md && \
  echo GREEN

Expected: prints GREEN.

• Step 4: Commit

git add .claude/agents/planner
git commit -m "iter pr.1.2: symlink .claude/agents/planner"

---

Task pr.1.3: Modify skills/planner/SKILL.md

Files:

• Modify: skills/planner/SKILL.md — replace Step 2 body; replace Cross-references "No private agents" line.

• Step 1: RED check — old Step 2 prose still present, new prose absent

Run:

grep -q "Before defining tasks, list every file" skills/planner/SKILL.md && \
  grep -q "No private agents" skills/planner/SKILL.md && \
  ! grep -q "Dispatch \`ailang-plan-recon\`" skills/planner/SKILL.md && \
  echo RED

Expected: prints RED.

• Step 2: Replace Step 2 body

Find the block (Step 2 header through the end of the closing sentence "Files that change independently get separate tasks."):

### Step 2 — Map file structure

Before defining tasks, list every file the iteration will create or
modify, and what each one is responsible for. This is where
decomposition decisions are locked in.

```markdown
**Files this plan creates or modifies:**

- Create: `crates/ailang-codegen/src/<new>.rs` — <one-line
  responsibility>
- Modify: `crates/ailang-check/src/typeclass.rs` — adds
  <responsibility>
- Test: `crates/ail/tests/e2e.rs` — adds <case>

Files that change together stay in the same task. Files that change independently get separate tasks.

Replace with (preserving the surrounding `### Step 3` heading
untouched):

```markdown
### Step 2 — Map file structure

Dispatch `ailang-plan-recon` with the spec path and iteration scope.
Read the returned file-map. Lift the relevant entries into the plan's
"Files this plan creates or modifies" section; the Boss may add or
amend entries that recon missed.

Dispatch carrier:

| Field | Content |
|-------|---------|
| `spec_path` | path to the parent spec |
| `iteration_scope` | which sections of the spec this iteration covers |
| `focus_hint` | optional — subsystem or symbol to prioritise |

Recon does NOT decide decomposition. Two recon dispatches per planner
run is unusual but legitimate (e.g. one for the main spec sections,
one with a sharper `focus_hint` for a tricky subsystem). Three or
more is a smell — re-read the spec; the iteration scope is probably
too broad.

The plan's "Files this plan creates or modifies" section keeps the
same shape:

```markdown
**Files this plan creates or modifies:**

- Create: `<exact path>` — <one-line responsibility>
- Modify: `<exact path>:<line range if known>`
- Test: `<exact path>` — <one-line case description>

Files that change together stay in the same task. Files that change independently get separate tasks.

- [ ] **Step 3: Replace Cross-references "No private agents" line**

Find the block:

```markdown
## Cross-references

- **Input source:** `skills/brainstorm/SKILL.md` — produces the
  spec this skill consumes.
- **Output target:** `skills/implement/SKILL.md` — runs the plan
  task-by-task.
- **No private agents.** This skill is dialogue-driven (orchestrator
  + spec).

Replace with:

## Cross-references

- **Input source:** `skills/brainstorm/SKILL.md` — produces the
  spec this skill consumes.
- **Output target:** `skills/implement/SKILL.md` — runs the plan
  task-by-task.
- **Agents dispatched:**
  - `skills/planner/agents/ailang-plan-recon.md` — Step 2,
    file-structure mapping (read-only). Mandatory; the Boss does
    NOT do recon in-context.

• Step 4: GREEN check

Run:

! grep -q "Before defining tasks, list every file" skills/planner/SKILL.md && \
  ! grep -q "No private agents" skills/planner/SKILL.md && \
  grep -q "Dispatch \`ailang-plan-recon\`" skills/planner/SKILL.md && \
  grep -q "Mandatory; the Boss does" skills/planner/SKILL.md && \
  echo GREEN

Expected: prints GREEN.

• Step 5: Commit

git add skills/planner/SKILL.md
git commit -m "iter pr.1.3: planner Step 2 dispatches ailang-plan-recon"

---

Task pr.1.4: Modify skills/brainstorm/SKILL.md

Files:

• Modify: skills/brainstorm/SKILL.md — extend Cross-references "No private agents" line.

• Step 1: RED check

Run:

grep -q "No private agents.\*\*\* This skill is dialogue-driven\." skills/brainstorm/SKILL.md && \
  ! grep -q "ad-hoc dispatch" skills/brainstorm/SKILL.md && \
  echo RED

Expected: prints RED.

• Step 2: Extend the "No private agents" bullet

Find the block:

## Cross-references

- **Project source:** `docs/DESIGN.md` "Feature-acceptance
  criterion" — gate this skill applies during spec writing.
- **Output target:** `skills/planner/SKILL.md` — only valid next
  skill.
- **No private agents.** This skill is dialogue-driven.

Replace with:

## Cross-references

- **Project source:** `docs/DESIGN.md` "Feature-acceptance
  criterion" — gate this skill applies during spec writing.
- **Output target:** `skills/planner/SKILL.md` — only valid next
  skill.
- **No private agents.** This skill is dialogue-driven. The Boss
  MAY ad-hoc dispatch `skills/planner/agents/ailang-plan-recon.md`
  during Step 1 when the milestone enters code territory the Boss
  has not recently read; this is opt-in and not part of the standard
  process. Rationale: brainstorm's recon phase is interwoven with
  the user Q&A — see `docs/specs/0009-plan-recon-subagent.md`
  for the full reasoning.

• Step 3: GREEN check

Run:

grep -q "ad-hoc dispatch" skills/brainstorm/SKILL.md && \
  grep -q "Rationale: brainstorm's recon phase is interwoven" skills/brainstorm/SKILL.md && \
  echo GREEN

Expected: prints GREEN.

• Step 4: Commit

git add skills/brainstorm/SKILL.md
git commit -m "iter pr.1.4: brainstorm allows ad-hoc plan-recon dispatch"

---

Task pr.1.5: Modify skills/README.md

Files:

• Modify: skills/README.md — insert agent roster row; extend the .claude/agents/ Discovery listing.

• Step 1: RED check

Run:

! grep -q "ailang-plan-recon" skills/README.md && \
  ! grep -q "\.claude/agents/planner" skills/README.md && \
  echo RED

Expected: prints RED.

• Step 2: Insert agent roster row at top of table

Find the line:

| `ailang-implement-orchestrator` | `implement/agents/` | `implement` (the only agent with the `Agent` tool; runs the per-task loop) |

Replace with two lines (the new entry first, then the existing implement-orchestrator line preserved):

| `ailang-plan-recon`     | `planner/agents/`   | `planner` (Step 2; read-only file-structure mapping); `brainstorm` (ad-hoc Step 1 in unfamiliar territory) |
| `ailang-implement-orchestrator` | `implement/agents/` | `implement` (the only agent with the `Agent` tool; runs the per-task loop) |

• Step 3: Extend the .claude/agents/ Discovery listing

Find the block:

.claude/agents/implement  -> skills/implement/agents
.claude/agents/audit      -> skills/audit/agents
.claude/agents/fieldtest  -> skills/fieldtest/agents
.claude/agents/debug      -> skills/debug/agents

Replace with (insert the planner line at top, matching pipeline order — planner is upstream of implement):

.claude/agents/planner    -> skills/planner/agents
.claude/agents/implement  -> skills/implement/agents
.claude/agents/audit      -> skills/audit/agents
.claude/agents/fieldtest  -> skills/fieldtest/agents
.claude/agents/debug      -> skills/debug/agents

• Step 4: GREEN check

Run:

grep -q "^| \`ailang-plan-recon\`" skills/README.md && \
  grep -q "^.claude/agents/planner" skills/README.md && \
  echo GREEN

Expected: prints GREEN.

• Step 5: Commit

git add skills/README.md
git commit -m "iter pr.1.5: roster + discovery entries for ailang-plan-recon"

---

Post-task checks (for the orchestrator-agent end-report)

• All five tasks committed.
• cargo build --workspace --release passes (sanity — should be unaffected, this iter touches no Rust).
• bench/check.py, bench/compile_check.py, bench/cross_lang.py return exit code 0 (audit step; not strictly required at end of iter, but trivial and a useful smoke).
• Per-iter journal docs/journals/2026-05-11-iter-pr.1.md written with the iter-or.1 template (Summary / Per-task subjects / Concerns / Known debt / Blocked detail / Commits / Stats / Audit close).
• docs/journals/INDEX.md gets a new pointer line.

---

Verification ladder (deferred — checked off when applicable)

These are the spec's acceptance-criteria items that cannot be verified within this iter itself. They get checked off in the iter journal as conditions arise:

1. First real planner dispatch uses ailang-plan-recon; the file-map names every path the plan ends up creating or modifying.
2. Token-budget delta at end of session: visible reduction compared with previous planner sessions (ct.1, ct.4, or.1).
3. Plan-quality preservation: the first plan written with the agent passes implement spec-review on Task 1 without non_compliant on stale paths.
4. BLOCKED handling exercise: when a recon dispatch organically blocks, the Boss handles via carrier-clarification.
5. Ad-hoc brainstorm dispatch: first brainstorm in unfamiliar territory after this lands tries the ad-hoc path.

---

Out of scope (mirrors spec)

• Brainstorm dedicated subagents (rationale: dialog-driven, recon interwoven with Q&A).
• Plan-writer subagent (rationale: decomposition IS the design work).
• Plan-reviewer subagent (rationale: today's inline self-review + downstream implement spec-review is sufficient gate).
• Modifying CLAUDE.md — none of this iter's changes update the orchestrator framing; the recon agent is one more agent under the existing skill system, not a new layer.