Step 9: deprecate subsumed specialists (v0.2 complete) (#14)

* Step 9: deprecate subsumed specialists (v0.2 complete)

Remove fetch-page-title and extract-structured-data from the advertised
skill set. Both are reachable via browser-task — page-title as a trivial
intent, structured extraction via a "return JSON: {...}" instruction
carried in the planner's done(result=...) channel. Cost delta is 2-3x
tokens per call, acceptable given zero deterministic high-volume
callers today. extract-structured-data was also out of spec on §7.1 —
it called the no-argument CreateSessionAsync overload and accepted any
host. The generalist enforces allowlists by design.

Advertised v0.2 surface lands at three skills: browser-task,
learn-form-schema, execute-form-batch.

- Delete FetchPageTitleCapability, ExtractStructuredDataCapability,
  and the shared CapabilityInput URL/description parser (no other
  consumers). browser-task has its own BrowserTaskInput; form
  capabilities have their own input classes.
- Delete the session-level one-shot helpers that only the removed
  specialists used: IBrowserSession.FetchPageTitleAsync,
  CapturePageSnapshotAsync, PageSnapshot, PageSnapshotSource.
- Delete the corresponding tests — 7 unit tests for the capabilities,
  the PlaywrightBrowserSessionTests + PageSnapshotTests integration
  suites, and the ExtractStructuredDataIntegrationTests real-LLM
  benchmark. BrowserTaskIntegrationTests remains the real-LLM surface.
- Trim StubBrowserSessionFactory + FakeAgentBrowserSession to match
  the pruned IBrowserSession.

Update metadata: deploy/rockbot-seed/*.json, docker-compose.yml
description + curl smoke example, .env.example comments, Program.cs
comment, docs/capabilities.md, spec §5.2 capability table and §9.1
step-9 description, CLAUDE.md Status + Browser + Capabilities
sections, framework-feedback step-9 section. Version bumped
0.2.0-alpha.8 → 0.2.0-alpha.9.

Tests: 48 passed (46 Agent unit + 1 Forms integration + 1 placeholder),
3 real-LLM BrowserTaskIntegrationTests skipped as expected. Build
clean on Release.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Step 9 follow-ups: fixes surfaced during end-to-end validation

Running RockBot → Foragent with a real browser-task (MacBook-price search
across apple.com + bestbuy.com) surfaced three pre-existing issues that
blocked step 9's claimed end-to-end validation. Fixing them here on the
step-9 branch keeps the PR's test plan honest.

1. BrowserTaskPriming required IEmbeddingGenerator (DI resolution bug)
   The primary-constructor parameter was annotated nullable
   (IEmbeddingGenerator<string, Embedding<float>>?), but MSDI ignores C#
   nullable annotations — it only honors default parameter values.
   Reordered to put embeddingGenerator last with = null so MSDI treats
   it as optional. Spec §5.6 says missing embeddings should downgrade
   to BM25-only retrieval; that claim is now actually true. Two test
   callers updated to drop the explicit embeddingGenerator: null arg.

2. Skill names with dotted hosts failed silently
   RockBot 0.9's FileSkillStore.ValidateName rejects '.' — every real
   host (bsky.app, apple.com, example.com) threw ArgumentException on
   save. BskySeedSkillService swallowed the throw as a startup warning,
   TryWriteLearnedSkillAsync swallowed it on the error path, and form
   schemas just never persisted. Added SkillNaming.SanitizeHost that
   replaces '.' → '-' (bsky.app → bsky-app) and applied it at three
   call sites: BskySeedSkillService, BrowserTaskCapability.
   TryWriteLearnedSkillAsync, LearnFormSchemaCapability.DeriveSkillName.
   Allowlist matching and memory-search categories keep the original
   dotted host — only skill names need sanitization. Test assertions
   (BrowserTaskCapabilityTests, BskySeedSkillServiceTests,
   LearnFormSchemaCapabilityTests) updated to the sanitized names;
   skill-optimize.md directive examples updated so the dream loop
   produces valid names.

3. Fresh named volume masks Dockerfile chown
   The Foragent Dockerfile chowns /data to the non-root foragent user
   (uid 1655) at image-build time, but Docker mounts a fresh named
   volume root-owned, masking the build-time chown. Added a
   foragent-init busybox one-shot (mirroring rockbot-init) that
   chmod -R 777 /data/foragent on volume creation.

Docs updated: CLAUDE.md Status + Learning-substrate sections,
docs/capabilities.md, spec §5.6 skill-naming paragraph (calls out the
sanitization rule), framework-feedback step-9 follow-up section with
three framework observations (MSDI nullable footgun, validator's dot
rejection making real hosts fail, named-volume permissions pattern).

Tests: 48 passed / 3 LLM-gated skipped. End-to-end smoke: RockBot
dispatches browser-task to Foragent over the bus; Foragent plans 2
steps (navigate + snapshot), emits done with JSON result, reply lands
on user.response.RockBot.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Step 9 follow-ups: A2A DataPart input + agent-card hints

When RockBot's LLM called invoke_agent with free-form prose (message=
"...allowedHosts: ['*']..."), Foragent's three input parsers kept
rejecting with 'Missing allowedHosts' — they only consumed text parts
and expected a JSON object. RockBot 0.9.11+ supports structured input
via an A2A DataPart (AgentMessagePart{Kind="data", Data=<json>}), but
Foragent never advertised that it consumed data parts and the invoke_agent
tool description steered the LLM to omit 'data' unless the target
"is known to consume data." Result: loop.

Fix spans three surfaces:

1. Parsers accept DataPart. BrowserTaskInput, LearnFormSchemaInput, and
   ExecuteFormBatchInput now look for a Kind="data" part first and use
   its Data string as the JSON source. Text-JSON fallback stays (curl
   callers), and for browser-task, a prose text part serves as the
   intent fallback when the data part doesn't supply one. Metadata
   overrides remain.

2. Skill descriptions explicitly direct callers to use the data
   parameter. Each SkillDefinition.Description now leads with "PASS
   INPUT AS AN A2A DATA PART (a structured JSON object), not as prose
   inside the text message. When calling via RockBot's invoke_agent,
   populate the 'data' parameter with this object." Matching entries
   in deploy/rockbot-seed/well-known-agents.json updated so the LLM
   sees the same guidance through list_known_agents.

3. Tests. Four new unit tests: one per input parser verifying a
   DataPart with JSON is consumed; one for browser-task's text-as-
   intent fallback when the data part omits intent. TestContext
   gained RequestWithData(...) to build the dual-part shape RockBot's
   invoke_agent produces.

Image bumped to rockylhotka/rockbot-agent:0.9.14 — softens the
invoke_agent 'data' tool description upstream, complementing the
skill-description hints on the Foragent side. CLAUDE.md Status
paragraph updated.

Docs: CLAUDE.md Capabilities section gains a note on the DataPart
contract. framework-feedback step-9 follow-up section extended with
the three-surface lesson (sender tool description ↔ target skill
description ↔ target parser all need to agree on the canonical shape).

Tests: 52 passed / 3 LLM-gated skipped. Build clean. Curl smoke
(text-JSON path) returns valid JSON via browser-task unchanged.
Live Blazor end-to-end test is next, against the updated 0.9.14
rockbot image.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Step 9 follow-ups: A2A task cancellation + self-teaching errors

RockBot 0.9.15 now publishes agent.task.cancel.{agentName} messages when
a wisp's local state fails after dispatching an A2A task (the duplicate-
dispatch scenario observed during step-9 validation). Foragent's
previous behavior inherited the framework's default
AgentTaskCancelHandler, which always replies TaskNotCancelable because
it assumes stateless agents. Foragent is stateful — a browser task is
potentially minutes long — so leaving the default would orphan browser
runs.

Implementation:
- InFlightTaskRegistry (singleton): ConcurrentDictionary<taskId,
  CancellationTokenSource> with Register/TryCancel/Remove. Register
  returns a linked CT that fires on either external cancel or the
  parent message CT. Redelivered task ids cancel the prior
  registration before replacing it, so stale work unwinds.
- ForagentTaskHandler wraps the capability's AgentTaskContext so the
  CT observed via context.MessageContext.CancellationToken is the
  linked one from the registry. Capabilities observe cancellation
  without any signature change.
- ForagentCancelHandler (IMessageHandler<AgentTaskCancelRequest>):
  on match calls TryCancel and publishes nothing (the running task's
  own terminal reply is the acknowledgment); on miss publishes
  AgentTaskError{Code=TaskNotFound}. Registered via
  agent.HandleMessage<AgentTaskCancelRequest, ForagentCancelHandler>()
  after AddA2A — last AddScoped wins, overriding the default.
- 11 new unit tests across registry, cancel handler, and task-handler
  integration (parent-cancel → linked CT fires, external cancel →
  linked CT fires, register/remove ties to finally, Remove drops
  registration even on thrown capability).

Also in this commit, incorporating earlier step-9 follow-ups for the
same RockBot 0.9.x interop round:

- Self-teaching errors. When a parser rejects for a missing required
  field, the response now tells the LLM exactly how to fix the call:
  "Pass inputs as a JSON object on the A2A DataPart — in RockBot's
  invoke_agent tool, that means filling the 'data' parameter, NOT
  adding fields to the 'message' text. Example data: {...}." Observed
  behavior: LLMs that ignore skill descriptions do read error replies
  and adjust subsequent calls. Applied to all three parsers.

- Docker image bumped to rockylhotka/rockbot-agent:0.9.15 — brings
  (a) invoke_agent's structured 'data' parameter (0.9.11), (b)
  softened tool description encouraging DataPart usage (0.9.14), and
  (c) the cancel-publisher that this commit consumes (0.9.15).
  CLAUDE.md Status section updated accordingly.

Framework-feedback step-9 follow-up section extended with the cancel-
handler-override pattern as a candidate for upstream WithTaskCancellation
ergonomics (non-blocking — ~50 LOC across consumers isn't unbearable).

Tests: 63 passed / 3 LLM-gated skipped. Build clean. Foragent starts
cleanly on fresh volumes; agent.task.cancel.Foragent subscription
verified active at boot.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: rockfordlhotka

.env.example

@@ -2,14 +2,13 @@
 #
 # These are consumed by docker-compose.yml. Foragent and the RockBot agent can
 # be pointed at different models — Foragent uses FORAGENT_LLM_* for its own
-# LLM-backed capabilities (extract-structured-data and beyond), the RockBot
-# agent uses LLM_* for its own reasoning.
+# LLM-backed capabilities (browser-task planner, form-schema enrichment), the
+# RockBot agent uses LLM_* for its own reasoning.
 #
-# Direct curl tests of capabilities that don't need an LLM (fetch-page-title)
-# still work without either set; anything LLM-backed will fail until Foragent's
-# config is populated.
+# Every Foragent capability needs the LLM wired; without it, the host fails
+# fast at startup.
 
-# ── Foragent's LLM (REQUIRED for extract-structured-data) ────────────────────
+# ── Foragent's LLM (REQUIRED — browser-task + form-schema enrichment) ────────
 # Azure AI Foundry / OpenAI-compatible endpoints are both fine.
 # Foundry endpoint shape: https://<resource>-<region>.cognitiveservices.azure.com/openai/v1/
 FORAGENT_LLM_ENDPOINT=https://your-resource-region.cognitiveservices.azure.com/openai/v1/

CLAUDE.md

@@ -6,7 +6,7 @@
     <ImplicitUsings>enable</ImplicitUsings>
     <TreatWarningsAsErrors>true</TreatWarningsAsErrors>
     <EnforceCodeStyleInBuild>true</EnforceCodeStyleInBuild>
-    <Version>0.2.0-alpha.8</Version>
+    <Version>0.2.0-alpha.9</Version>
     <Authors>Marimer LLC</Authors>
     <Company>Marimer LLC</Company>
     <Copyright>Copyright (c) Marimer LLC</Copyright>

Directory.Build.props

@@ -2,7 +2,7 @@
   {
     "agentId": "Foragent",
     "level": 4,
-    "approvedSkills": ["browser-task", "fetch-page-title", "extract-structured-data"],
+    "approvedSkills": ["browser-task", "learn-form-schema", "execute-form-batch"],
     "firstSeen": "2026-04-21T00:00:00+00:00",
     "lastInteraction": "2026-04-21T00:00:00+00:00",
     "interactionCount": 0

deploy/rockbot-seed/agent-trust.json

@@ -1,7 +1,7 @@
 [
   {
     "agentName": "Foragent",
-    "description": "Browser agent — navigates pages with Chromium and exposes task-level skills over HTTP A2A.",
+    "description": "Browser agent — navigates pages with Chromium and exposes task-level skills over HTTP A2A. All skills consume structured input on the A2A DataPart (invoke_agent 'data' parameter); the text 'message' is only for human-readable summaries.",
     "version": "1.0",
     "url": "http://foragent:8080",
     "protocolVersion": "1.0",
@@ -11,17 +11,17 @@
       {
         "id": "browser-task",
         "name": "Browser Task (generalist)",
-        "description": "Drive a browser with an LLM-in-the-loop planner to accomplish a free-form intent. Input JSON {\"intent\":\"...\",\"allowedHosts\":[\"host\",\"*.host\",\"*\"],\"url\":\"optional start\",\"credentialId\":\"optional\",\"maxSteps\":60,\"maxSeconds\":120}. allowedHosts is required and empty rejects. Returns a structured JSON result with status (done/failed/incomplete), summary, optional result, step count, and navigations."
+        "description": "Drive a browser with an LLM-in-the-loop planner to accomplish a free-form intent. PASS INPUT AS AN A2A DATA PART — populate invoke_agent's 'data' parameter with {\"intent\":\"...\",\"allowedHosts\":[\"host\",\"*.host\",\"*\"],\"url\":\"optional start\",\"credentialId\":\"optional\",\"maxSteps\":60,\"maxSeconds\":120}. 'intent' and 'allowedHosts' are required; an empty allowlist rejects the task. Use [\"*\"] explicitly when any host is acceptable. Returns a structured JSON result with status (done/failed/incomplete), summary, optional result, step count, and navigations."
       },
       {
-        "id": "fetch-page-title",
-        "name": "Fetch Page Title",
-        "description": "Navigate to a URL with a real browser and return the contents of its <title> element."
+        "id": "learn-form-schema",
+        "name": "Learn Form Schema",
+        "description": "Navigate to a web form, extract its structure (fields, types, options, validation), and persist it as a reusable skill. PASS INPUT AS AN A2A DATA PART — populate invoke_agent's 'data' parameter with {\"url\":\"https://...\",\"allowedHosts\":[\"host\"],\"formSelector\":\"optional\",\"credentialId\":\"optional\",\"skillName\":\"optional override\",\"intent\":\"optional prose\"}. 'url' and 'allowedHosts' are required. Returns the typed form schema plus the skill name it was persisted under."
       },
       {
-        "id": "extract-structured-data",
-        "name": "Extract Structured Data",
-        "description": "Navigate to a URL and extract data matching a natural-language description, returning JSON. Input the target URL and a description of what to extract."
+        "id": "execute-form-batch",
+        "name": "Execute Form Batch",
+        "description": "Submit a batch of rows against a learned form schema. PASS INPUT AS AN A2A DATA PART — populate invoke_agent's 'data' parameter with {\"schemaRef\":\"sites/host/forms/name\" OR \"schema\":{...FormSchema...},\"rows\":[{fieldName:value,...}],\"allowedHosts\":[\"host\"],\"credentialId\":\"optional\",\"mode\":\"abort-on-first\"|\"continue\",\"successIndicator\":\"optional CSS selector\"}. 'rows', 'allowedHosts', and exactly one of schemaRef/schema are required. Streams per-row progress. Default mode aborts on first failure."
       }
     ]
   }

deploy/rockbot-seed/well-known-agents.json

@@ -5,7 +5,7 @@
 #   - foragent        — this project; exposes HTTP A2A on port 5210
 #   - rockbot-init    — seeds /data/agent with RockBot profile + well-known-agents.json
 #                       pointing at foragent
-#   - rockbot         — rockylhotka/rockbot-agent:0.9.11, configured to know Foragent
+#   - rockbot         — rockylhotka/rockbot-agent:0.9.15, configured to know Foragent
 #                       as an A2A peer it can delegate tasks to. 0.9.11 brings
 #                       the structured-data invoke_agent surface (PR #291) so
 #                       RockBot can consume Foragent's FormSchema JSON results
@@ -21,7 +21,7 @@
 #   curl -X POST http://localhost:5210/ \
 #     -H "X-Api-Key: rockbot-calls-foragent" \
 #     -H "Content-Type: application/json" \
-#     -d '{"jsonrpc":"2.0","id":1,"method":"message/send","params":{"message":{"role":"ROLE_USER","messageId":"m1","parts":[{"text":"https://example.com"}]},"metadata":{"skill":"fetch-page-title"}}}'
+#     -d '{"jsonrpc":"2.0","id":1,"method":"message/send","params":{"message":{"role":"ROLE_USER","messageId":"m1","parts":[{"text":"{\"intent\":\"fetch the page title\",\"url\":\"https://example.com\",\"allowedHosts\":[\"example.com\"]}"}]},"metadata":{"skill":"browser-task"}}}'
 # Note: the A2A v1-preview schema uses protobuf-style enum values (ROLE_USER, not "user")
 # and parts are bare {"text":"..."} objects — no "kind" field.
 #
@@ -47,12 +47,27 @@ services:
       timeout: 3s
       retries: 15
 
+  foragent-init:
+    # One-shot ownership fix for the foragent-data named volume. The Foragent
+    # Dockerfile chowns /data to the non-root `foragent` user at build time, but
+    # Docker mounts a fresh named volume root-owned and masks the chown —
+    # FileSkillStore then hits UnauthorizedAccessException on first boot. This
+    # init container runs as root once per volume creation and mirrors the
+    # rockbot-init pattern. Subsequent boots skip the mkdir+chown if already set.
+    image: busybox:latest
+    user: root
+    command: ["sh", "-c", "mkdir -p /data/foragent/skills /data/foragent/memory && chmod -R 777 /data/foragent"]
+    volumes:
+      - foragent-data:/data/foragent
+
   foragent:
     build:
       context: .
     depends_on:
       rabbitmq:
         condition: service_healthy
+      foragent-init:
+        condition: service_completed_successfully
     ports:
       - "5210:8080"
     environment:
@@ -64,12 +79,12 @@ services:
       RabbitMq__VirtualHost: /
       Gateway__AgentName: Foragent
       Gateway__InternalAgentName: Foragent
-      Gateway__Description: "Browser agent — browser-task (generalist), learn-form-schema, execute-form-batch, fetch-page-title, extract-structured-data"
+      Gateway__Description: "Browser agent — browser-task (generalist), learn-form-schema, execute-form-batch"
       # RockBot will call Foragent with header X-Api-Key: rockbot-calls-foragent
       ApiKeys__rockbot-calls-foragent__AgentId: RockBot
       ApiKeys__rockbot-calls-foragent__DisplayName: RockBot
-      # LLM required for the extract-structured-data capability. Namespaced so
-      # Foragent can point at a different model than the RockBot side.
+      # LLM required for the browser-task planner and form-schema enrichment.
+      # Namespaced so Foragent can point at a different model than the RockBot side.
       ForagentLlm__Endpoint: ${FORAGENT_LLM_ENDPOINT:?FORAGENT_LLM_ENDPOINT is required}
       ForagentLlm__ModelId: ${FORAGENT_LLM_MODEL_ID:?FORAGENT_LLM_MODEL_ID is required}
       ForagentLlm__ApiKey: ${FORAGENT_LLM_API_KEY:?FORAGENT_LLM_API_KEY is required}
@@ -100,7 +115,7 @@ services:
       - foragent-data:/data/foragent
 
   rockbot-init:
-    image: rockylhotka/rockbot-agent:0.9.11
+    image: rockylhotka/rockbot-agent:0.9.15
     user: root
     entrypoint: ["/bin/sh", "-c"]
     command:
@@ -136,7 +151,7 @@ services:
       - ./deploy/rockbot-seed:/seed:ro
 
   rockbot:
-    image: rockylhotka/rockbot-agent:0.9.11
+    image: rockylhotka/rockbot-agent:0.9.15
     depends_on:
       rockbot-init:
         condition: service_completed_successfully

docker-compose.yml

@@ -8,13 +8,22 @@ invoke capabilities by name; Foragent handles the browser mechanics.
 - `browser-task` — **generalist**, spec §5.2. LLM-in-the-loop planner that
   drives a real browser to accomplish a free-form intent. Shipped in
   step 6; step 7 added skills + memory priming (spec §5.6).
-- `fetch-page-title` — specialist. Inherited from step 1/2.
-- `extract-structured-data` — specialist. Inherited from step 3.
-
-The step-4 `post-to-site` capability was removed in step 7 — the
-generalist `browser-task` plus the seeded `sites/bsky.app/login` skill
-subsume its function, and the project is still pre-public so no consumer
-needed a deprecation path.
+- `learn-form-schema` — specialist (phase-1, spec §5.5). Introspects a
+  form and returns a typed `FormSchema` persisted at `sites/{host}/forms/{slug}`.
+- `execute-form-batch` — specialist (phase-3, spec §5.5). Submits rows
+  against a learned schema, streaming per-row progress over A2A.
+
+Three v0.1/v0.2 specialists have been removed as `browser-task` subsumes
+them. The project is pre-public so no deprecation path was required:
+
+- `post-to-site` — removed in step 7. `browser-task` + the seeded
+  `sites/bsky-app/login` skill cover the use case.
+- `fetch-page-title` — removed in step 9. Was a milestone-1 smoke
+  target; `browser-task` with a simple intent produces the same result.
+- `extract-structured-data` — removed in step 9. `browser-task` with a
+  "return JSON: {…}" intent produces the same result. Its typed input
+  shape also lacked the mandatory allowlist required by spec §7.1; the
+  generalist enforces that by design.
 
 ## `browser-task` input shape
 
@@ -60,7 +69,12 @@ A JSON object in a single text part:
 ```
 
 `incomplete` means the budget was exhausted before `done`/`fail` was
-called.
+called. For extraction-style tasks, instruct the planner to return JSON
+via the `result` field — e.g. intent `"Open https://shop.example/p/42
+and return {\"name\":..., \"price_usd\":...} as JSON in the result
+field."`. The planner is not schema-enforced the way
+`extract-structured-data` used to be, so keep the target shape explicit
+in the intent.
 
 ## `browser-task` tool surface
 

docs/capabilities.md

@@ -311,15 +311,15 @@ callers cheap — not to proliferate.
 | `browser-task` | Generalist | Given intent + optional URL, credential id, and allowed-hosts list, plan and drive the browser to fulfill the intent. Uses RockBot skills + memory as priming. Returns a result or a structured intermediate artifact (e.g. a learned form schema). |
 | `learn-form-schema` | Specialist (phase-1) | Given a URL and optional credential, introspect a form and return its schema — fields, types, dropdown dependencies, validation rules. Persists the schema as a skill (§5.6). Returns the schema to the caller for review. |
 | `execute-form-batch` | Specialist (phase-2) | Given a learned schema (by id or inline) and a batch of row data, submit the form once per row. Streams A2A progress updates. Handles partial failure. |
-| `fetch-page-title` | Specialist | Return the `<title>` of a URL. Inherited from milestone 2. |
-| `extract-structured-data` | Specialist | Extract structured data from a page matching a natural-language description. Inherited from milestone 3. |
 
-The v0.1 `post-to-site` capability ships in the main codebase as a
-regression test for credential handling. After step 7 it is removed
-from the advertised skill list; `browser-task` subsumes its function.
-
-The v0.1 `monitor-page` and `fill-form` capabilities fold into
-`browser-task` and do not ship as separate advertised skills.
+After step 9 the v0.2 surface is three skills. The v0.1 `post-to-site`
+capability was removed in step 7 once the seeded `bsky.app` skill +
+`browser-task` covered it; the v0.1 `fetch-page-title` and v0.1
+`extract-structured-data` specialists were removed in step 9 — the
+generalist subsumes both at the cost of 2–3× tokens per call, which
+is acceptable given zero deterministic high-volume callers today. The
+v0.1 `monitor-page` and `fill-form` capabilities fold into `browser-task`
+and do not ship as separate advertised skills either.
 
 ### 5.3 Capabilities explicitly out of scope (v1)
 
@@ -393,18 +393,21 @@ site knowledge, rather than building a Foragent-local store.
   `RockBot.Host.Abstractions` + `RockBot.Host.AgentMemoryExtensions.WithSkills()`).
   Stores site knowledge as markdown skills. Two origin categories:
   - **Human-authored skills** — operator-written primers for a site
-    (e.g. `sites/bsky.app/overview`). Treated as priming hints for the
+    (e.g. `sites/bsky-app/overview`). Treated as priming hints for the
     generalist planner.
   - **Agent-learned skills** — written by the generalist on successful
-    task completion (e.g. `sites/bsky.app/learned/login-flow`). Tagged
+    task completion (e.g. `sites/bsky-app/learned/login-flow`). Tagged
     with `metadata.source = "agent-learned"` and an importance score.
 - **`ILongTermMemory`** (file-backed, BM25 + semantic —
   `WithLongTermMemory()`). Declarative observations that don't fit the
   procedural skill shape: failed attempts, site-version notes, ambient
   facts.
 
 **Skill naming:** `sites/{host}/{phase-or-intent}` — e.g.
-`sites/bsky.app/login`, `sites/bsky.app/compose-post`. Hierarchical `/`
+`sites/bsky-app/login`, `sites/bsky-app/compose-post`. Host segments are
+sanitized (`.` → `-`) because RockBot 0.9's `FileSkillStore.ValidateName`
+rejects dots; `bsky.app` becomes `bsky-app`. Allowlists and memory
+categories keep the original dotted host. Hierarchical `/`
 nesting is supported by the store. `seeAlso` links cross-reference
 skills for the same site so retrieval surfaces a small knowledge
 cluster, not one skill at a time.
@@ -714,10 +717,15 @@ hard design questions until usage forces them.
    Resolve open question #6 (how to persist typed JSON alongside
    markdown skills) in the deliverable.
 
-9. **Deprecate subsumed specialists.** Review whether `fetch-page-title`
-   / `extract-structured-data` still pay their way or fold into
-   `browser-task` with equivalent cost. Land on the minimum advertised
-   capability set v0.2 actually needs.
+9. **Deprecate subsumed specialists.** Reviewed whether `fetch-page-title`
+   / `extract-structured-data` still paid their way vs. `browser-task` at
+   equivalent cost. Both removed: `fetch-page-title` was a milestone-1
+   smoke-test relic that `browser-task` subsumes trivially;
+   `extract-structured-data` was functionally equivalent to a `browser-task`
+   intent that asks for JSON in the `done.result` channel (cost delta
+   ~2–3× tokens per call, zero deterministic high-volume callers today),
+   and was out of spec on §7.1 mandatory allowlists. Advertised surface
+   lands at `browser-task` + `learn-form-schema` + `execute-form-batch`.
 
 Each milestone produces framework feedback. Capture it in
 `docs/framework-feedback.md` — some will be small ergonomic fixes; some