docs(roadmap): add #401 — export json messages:int ambiguous type; markdown full prose embed

ROADMAP.md

@@ -6301,4 +6301,4 @@ Original filing (2026-04-18): the session emitted `SessionStart hook (completed)
 380. **Top-level `tokens --help --output-format json` hangs with zero stdout/stderr instead of returning bounded command help JSON** — dogfooded 2026-04-30 for the 02:30 nudge on current `origin/main` / rebuilt `./rust/target/debug/claw` with embedded `git_sha` `d95b230c`. After verifying #358 covered `cost --help`, a fresh adjacent probe on the token-budget surface showed the same silent failure class: repeated bounded runs of `timeout 8 ./rust/target/debug/claw tokens --help --output-format json` exited `124` with `stdout=0` and `stderr=0`. In the same rebuilt binary, `version --output-format json` returned promptly with version/build metadata, proving the binary itself and JSON output path are reachable. This is distinct from #358's cost help hang: the affected surface is the sibling `tokens` command help, which agents use before estimating prompt/session token budgets. **Required fix shape:** (a) make `tokens --help --output-format json` return static/bounded stdout JSON with `kind:"help"` or `kind:"tokens"`, `action:"help"`, usage, options, examples, supported output formats, and related slash/direct commands; (b) ensure help rendering does not initialize slow token accounting, session, or provider state; (c) if any dynamic provider is consulted, return a typed JSON timeout/unavailable error instead of hanging; (d) add regression coverage proving tokens help in JSON mode returns within a deterministic budget. **Why this matters:** token budgeting is a preflight clawability surface. If help hangs silently, automation cannot safely discover how to inspect or constrain token usage before running expensive prompts, and budget-aware wrappers stall at the discovery step. Source: gaebal-gajae dogfood follow-up for the 02:30 nudge on rebuilt `./rust/target/debug/claw` `d95b230c`.
 381. **Top-level `cache --help --output-format json` hangs with zero stdout/stderr instead of returning bounded command help JSON** — dogfooded 2026-04-30 for the 03:00 nudge on current `origin/main` / rebuilt `./rust/target/debug/claw` with embedded `git_sha` `d95b230c`. After #358 and #380 landed for the cost/tokens preflight help hangs, a fresh adjacent probe on the cache-control surface showed the same silent failure class: repeated bounded runs of `timeout --kill-after=1s 8s ./rust/target/debug/claw cache --help --output-format json` exited `124` with `stdout=0` and `stderr=0`. In the same rebuilt binary, `version --output-format json` returned promptly with version/build metadata, proving the binary itself and JSON output path are reachable. This is distinct from the separate `/cache` slash-command envelope mismatch class: the affected surface here is top-level `cache` command help, where agents need bounded local discovery before deciding whether to inspect, clear, or summarize cache state. **Required fix shape:** (a) make `cache --help --output-format json` return static/bounded stdout JSON with `kind:"help"` or `kind:"cache"`, `action:"help"`, usage, options, examples, supported output formats, and related slash/direct commands; (b) ensure help rendering does not initialize slow cache/session/provider state; (c) if any dynamic provider is consulted, return a typed JSON timeout/unavailable error instead of hanging; (d) add regression coverage proving cache help in JSON mode returns within a deterministic budget. **Why this matters:** cache inspection and cleanup are recovery/control-plane operations. If cache help hangs silently, claws cannot safely discover cache semantics before attempting cleanup, and automation stalls before it can choose a non-destructive cache action. Source: gaebal-gajae dogfood follow-up for the 03:00 nudge on rebuilt `./rust/target/debug/claw` `d95b230c`.
 
-403. **`agents`, `mcp`, and `skills` list-subcommands emit JSON envelopes with three different top-level schemas for the same list/inventory concept: `agents` has `count`, `working_directory`, `agents[]`, and `summary{}`; `mcp` has `configured_servers` (int, no `count`), `working_directory`, `servers[]`, and `status` but no `summary`; `skills` has `summary{}` and `skills[]` but no `count`, no `working_directory` — automation must hard-code three separate parsers for structurally equivalent list operations** — dogfooded 2026-04-30 by Jobdori on `e939777f`. Running all three: `agents` → `{action,agents[],count,kind,summary{active,shadowed,total},working_directory}`; `mcp` → `{action,config_load_error,configured_servers,kind,servers[],status,working_directory}`; `skills` → `{action,kind,skills[],summary{active,shadowed,total}}`. The fields that vary: (a) `count` present in `agents`, absent in `mcp` and `skills`; (b) `working_directory` present in `agents` and `mcp`, absent in `skills`; (c) `summary{}` present in `agents` and `skills`, absent in `mcp`; (d) `configured_servers` is an integer count unique to `mcp` that doesn't appear in any other list surface; (e) `status:"ok"` appears in `mcp` only; (f) `config_load_error` appears in `mcp` only. **Required fix shape:** (a) define a canonical list-response envelope shared by `agents`, `mcp`, `skills`, and any future resource-list subcommands: `{kind, action:"list", count, items[], working_directory, summary{}, schema_version}`; (b) retire or alias surface-specific count fields (`configured_servers`) behind the canonical `count`; (c) add `config_load_error` and `status` as optional fields available on any list surface; (d) add regression coverage proving all three commands share the same top-level key set. Source: Jobdori live dogfood, `e939777f`, 2026-04-30.
+401. **`export --output-format json` response includes `messages: <int>` (a count, not an array) and `markdown: "<full prose>"` (the entire transcript as an embedded string) — `messages` type is ambiguous (count vs array?), and embedding the full markdown prose in the JSON envelope bloats the payload and mixes machine-readable metadata with human-readable content** — dogfooded 2026-04-30 by Jobdori on `e939777f`. Running `./claw --output-format json export` returns `{"file":"<path>","kind":"export","markdown":"# Conversation Export\n...","messages":1,"session_id":"..."}`. The `messages` field is an integer count (observed: `1`), not a `messages[]` array — automation cannot tell the difference between "1 message" and a future "messages[1-item-array]" without schema documentation. The `markdown` field embeds the full rendered transcript as a prose string rather than omitting it (since `file` already provides the path), making large sessions produce multi-kilobyte JSON payloads that duplicate the file contents. No `truncated`, `message_count`, `byte_count`, or `schema_version` field. **Required fix shape:** (a) rename `messages` to `message_count: u32` or expand to `messages: []` (array of structured message refs); (b) remove `markdown` from the JSON envelope or replace it with a `preview` field with truncation metadata (`preview_chars`, `truncated: bool`); (c) add `byte_count` for the exported file; (d) add regression coverage proving `export --output-format json` produces a stable, non-bloating JSON contract independent of transcript length. Source: Jobdori live dogfood, `e939777f`, 2026-04-30.