* chore: land the #2247 round-9 review findings on v2/main
Copilot filed a ninth review round on #2247 after it had merged, so none of
it reached v2/main. All five findings are addressed here.
1. The testing -> test-servers hand-off delivered the wrong half of the
procedure. test-servers documented only the two-process manual path
(server-composable.js --config <name>.json plus a second Inspector
process), but an integration test builds the server in-process from the
@modelcontextprotocol/inspector-test-server API and owns its lifecycle.
Adds a "Two ways to use a fixture — pick the right one first" section
with the in-process recipe, the four mechanics that bite (one barrel,
start/stop in afterEach, read started.url, era is a constructor option),
and a pointer to the reference test; renames "Run one" to "Run one by
hand (two processes)" so the manual path names itself. testing now says
in the connecting bullet that an automated test does NOT spawn.
2. testing's heading asked "does the test need a real server?", which its
own hand-rolled-server exception contradicts. It now asks whether the
test uses a test-servers/ fixture.
3. "imports" was too broad a dependency test — test-server-scope.test.ts
imports the package to unit-test scope middleware as a pure function and
never starts a server. The condition is now "starts, spawns, or
configures", with that test named as the counterexample.
4. docs/skill-authoring.md attributed the 33% -> 100% rise to imperative
wording and placement independently. The #2247 edit changed both at
once, so only the combination is established; the split into two levers
is now labelled the plausible mechanism, not a measured result.
5. The CHAIN_THRESHOLD comment generalised "a well-shaped pointer clears
0.8" from one two-case run of that same multi-variable edit. It now
reads as an existence proof about the ceiling, not a property of
pointers in general.
Also records the headline lesson where it can be found again: a green
hand-off number proves skill B loaded, never that B answered the question —
in AGENTS.md's chain-case rules and in the skill-authoring checklist.
Docs, skill bodies and one comment only; no runtime code changes.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018zwYWanPhvgu6wiKoeGg4N
Signed-off-by: cliffhall <cliff@futurescale.com>
* fix: address Copilot review round 1 on #2264
The new section claimed two shapes and there are three — the automated
stdio path was missing, and it is the repo's established pattern
(getTestMcpServerCommand() into a stdio transport, used by five web
integration files and the whole CLI suite). Telling callers an automated
test "does not spawn" told them not to use it.
- test-servers: "Two ways" -> "Three ways", with a table and a subsection
for the spawned-stdio shape — the transport owns the child (disconnect,
not stop()), no config can be selected because that entry point runs the
stdio default, and it is still the build. Names a reference test for it.
- test-servers: the build is no longer scoped to the manual path. All three
shapes resolve test-servers/build/ — two through the barrel alias, one by
running the emitted .js — so the staleness hazard is called out as
applying to every shape, and `Build first` no longer opens by claiming
the only reason is subprocess spawning.
- testing: the "does not spawn" instruction is replaced by the two
automated shapes and when to pick each. What they share, and what the
hand-off actually needed to say, is that neither picks a showcase config.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018zwYWanPhvgu6wiKoeGg4N
Signed-off-by: cliffhall <cliff@futurescale.com>
* fix: address Copilot review round 2 on #2264
Round 1 fixed the missing stdio shape but kept the wrong axis. Cutting
the taxonomy by "automated vs by hand" excluded the config-driven smokes
— smoke-web-elicitation.mjs spawns server-composable.js --config
directly, and lib/mcp-app-flow.mjs does it for smoke:web:app,
smoke:web:tabs and pack:verify. Those are automated AND need every bit
of the config and protocol-era guidance a person at two terminals does,
so the skill was routing their authors away from it.
- test-servers: the cut is now how the SERVER IS STOOD UP, stated as
such. Third shape is "spawned composable HTTP", whose consumers are
the smokes, pack:verify, and a person by hand — the config/era
guidance belongs to the shape, whoever starts it. Table gains a
Consumers column and loses the Who column that caused this.
- test-servers: the stale "an automated test builds the server
in-process instead" line under `Run one by hand` is gone (suppressed
comment); that heading is now the same shape driven by a person.
- testing: routes by what is driving — integration/CLI test to
in-process HTTP, stdio-is-the-point to spawned stdio, and a smoke to
the composable-config path with the era guidance it actually needs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018zwYWanPhvgu6wiKoeGg4N
Signed-off-by: cliffhall <cliff@futurescale.com>
* fix: address Copilot review round 3 on #2264
Round 2 replaced one wrong axis with a narrower one: it routed "a smoke"
wholesale to the config-driven composable path, and smoke:cli does not
use it. smoke-cli.mjs starts an in-process createTestServerHttp for the
header round-trip (so it can read back what the CLI sent) and writes the
built stdio entry into a --catalog for the connect checks; smoke:tui uses
only the latter. Neither picks a config.
- test-servers: consumers filled in per row — in-process HTTP gains CLI
tests and smoke:cli, spawned stdio gains smoke:cli and smoke:tui, and
the composable row is narrowed to the WEB smokes. A warning states the
point directly: "a smoke" is not a shape, pick by the row.
- test-servers: the stdio shape now covers naming the built entry in a
catalog without connecting, which is what those two smokes do.
- testing: same narrowing, plus the reason spelled out so the word
"smoke" stops being a routing key.
- testing: the dependency criterion said "starts, spawns, or configures",
which excluded the build-only smoke:tui case documented directly above
it — the skill both required and rejected the hand-off for it
(suppressed comment). It now reads "starts, spawns, configures, or
hands a built entry to the subject under test".
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018zwYWanPhvgu6wiKoeGg4N
Signed-off-by: cliffhall <cliff@futurescale.com>
---------
Signed-off-by: cliffhall <cliff@futurescale.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
- The recovery text told users to move the endpoint to `::1`. A bare IPv6
literal is not a legal URL host — `new URL("http://::1/token")` throws — so
anyone copying that into the Token URL override got a parse error from advice
meant to unblock them. Now `[::1]`, in both the notice and the manual
reproduction guide.
The exemption list still reads `::1`, because that is the host the SDK
compares; only the remedy is bracketed, because that is what a user types.
The comment says so, so neither gets "corrected" into the other.
- Dropped SEP-2207 from all nine files it had reached. I took the label from
the SDK's own source comments, which attribute this check to it — but this
repo already uses SEP-2207 for OIDC refresh / offline_access
(specification/v2_auth_hardening.md, plus an e2e test), so grepping it here
would have turned up two unrelated things. Replaced with plain description
rather than a different SEP number: having guessed wrong once, guessing again
is not the fix.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018MvNbmAtQuWhDvzi4yXDju
Signed-off-by: cliffhall <cliff@futurescale.com>
Closes#2280
InsecureTokenEndpointError appeared nowhere in our source, so it fell through
to the generic auth-failure path and rendered a "Re-authentication required"
banner with a Re-authenticate button. That button can never work: the SDK's
assertSecureTokenEndpoint runs inside executeTokenRequest, the error does not
extend OAuthError, and auth() special-cases it to rethrow rather than start a
fresh /authorize redirect. Clicking re-ran the same flow to the same refusal,
under the raw SDK message, which names the three exempt host literals and
nothing actionable. A terminal configuration error was being presented as a
retryable auth error.
Not specific to *.localhost: it fires for any endpoint outside the SDK's
exemption — host.docker.internal (#1911), a LAN hostname, a reverse-proxy
name, a mistyped scheme.
It is now recognized (core/auth/insecureTokenEndpoint.ts, mirroring
issuerBinding.ts's brand-plus-name classifier and walking cause / data.cause,
since era negotiation and the transport wrappers bury the rejection) and
surfaced as the configuration error it is, naming the endpoint and both ways
out, with no action affordance.
Six paths reach the refusal and all six classify it — the connect handshake,
authenticate() during connect, the satisfied-challenge connect retry, the
post-redirect callback, the command/background path, the deferred tab-visible
resume (which re-armed on every focus, an unbounded loop on a terminal error),
and the banner action. They share one wrapper so a seventh cannot omit the
banner clear, and every clear is scoped to its own serverId: these paths are
asynchronous, so a late continuation for one server must not erase a banner
another raised in the meantime.
Adds oauth-insecure-token-endpoint-http.json to reproduce it, whose issuer is
http://localhost.:8091 — the root-anchored spelling every resolver sends to
loopback but which is none of the SDK's exempt literals. transport.strictPort
keeps that fixture honest, since its port is hard-coded inside the issuer.
Making such an endpoint work has to land in the SDK (typescript-sdk#2591);
this changes the reporting, not the outcome.
Split out of #2282, where it was tangled with the *.localhost origin work for
#1944 — a separate question the reporter is still clarifying.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018MvNbmAtQuWhDvzi4yXDju
Signed-off-by: cliffhall <cliff@futurescale.com>
Five findings, all real.
**The chain prompts carried the target skill's own trigger.** `test-servers`
claims "a change needs a real server to exercise it", and both cases said
"against a real/live server" — so the model could load `testing`, then pick
`test-servers` from the ORIGINAL prompt, in that order, and score a hit that
would have survived deleting the pointer from `testing` entirely. Rewritten to
carry no server cue ("end to end"), and the measured rate fell from 100% / 67%
to 33% / 33%. That is the size of the artefact, and 33% is what the pointer
alone is worth. Both cases are now red against the 0.5 bar and stay that way:
`skills:eval` is not a gate, the number IS the finding, and lowering the bar to
turn the column green would discard the only signal this feature adds. Filed as
#2247 with the trap written down so a fix cannot re-measure the artefact.
**A deny list cannot bound a 14-turn run.** It only names tools known when it
was written, and this checkout configures an HTTP `mcp-docs` server in
`.mcp.json` while a contributor's own MCP servers and plugins add more. Now
`--allowedTools Read,Glob,Grep,Skill` (what the harness actually needs) plus
`--strict-mcp-config` with no `--mcp-config`, which drops every configured
server. The deny list stays alongside, since a deny is unconditional while an
allow list governs pre-approval. What remains outside all three is a
contributor's plugin tools; `--bare` would remove those and skills with them,
which would measure nothing — so that residual is stated rather than papered
over.
**The chain threshold compared inclusively.** "More often than not" is `> 0.5`,
and `>=` passes exactly half whenever RUNS is even — 2/4 would report a result
the criterion does not license. `passesThreshold` is now strict for chains and a
floor for first moves, and a strict bound of 1.0 is rejected up front rather
than failing every case while looking like a trigger problem.
**The doc's probe command still denied only four tools**, contradicting the
section's own promise that a probe and a scored case see the same policy.
Updated, with a note that it must move with the harness, plus a hand-off probe
form.
**The reporting had no automated coverage** — the one thing that could silently
merge the two measurements lived inside `main`. Extracted as `formatReport` and
tested: both headings, per-group thresholds, the two separate summaries, the
combined exit, and both single-kind selections.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ABdTnqyBodw6mHVVkGehqe
Signed-off-by: cliffhall <cliff@futurescale.com>
`skills:eval` ran every case with `--max-turns 1`, so it measured one
thing: whether a skill is the model's FIRST tool call. That is right for a
skill a user reaches directly, and it leaves a whole class unmeasured — a
skill reached from inside another skill's body. `testing` tells the model
that picking a fixture is `/test-servers` and that it has to load it, and
`test-servers` scores 5/5 while every one of those five cases asks for it
by name. A skill only ever reachable through a hand-off would score a
clean 100% with the hand-off silently never firing.
An eval case is now one of two shapes, and exactly one:
{ "prompt": "…", "expect": "test-servers" } first move, 1 turn
{ "prompt": "…", "chain": ["testing", "test-servers"] } hand-off, 14 turns
A chain must END with the skill whose file it lives in. The case exists to
measure whether THIS skill is reachable, so the file that goes red is the
one belonging to the skill that stopped being reached; anchoring on the
first link would file a `test-servers` measurement under `testing`, where
a `test-servers` description edit would never be seen. Links are checked
against the repo's model-invoked set, which is why both `verify:skills`
and `collectCases` now validate in a second pass — `test-servers` sorts
before `testing`, so a per-directory check would reject a live chain as
unknown purely because of where the alphabet put it.
Scoring is an ordered SUBSEQUENCE, not a prefix and not a contiguous run:
the model may load something before the chain starts and something
unrelated in between, and neither changes the claim that A led to B.
`collectSkillInvocations` therefore returns an ordered array rather than a
Set — a B, A, B run has to stay distinguishable from one that never
reached B from A.
The two rates are reported in separate columns and never summed. A
hand-off is a second-hop load over many turns; a first-move rate is the
model's opening move. `CHAIN_MAX_TURNS` (14) and `CHAIN_THRESHOLD` are
their own knobs for the same reason, and `CHAIN_THRESHOLD` defaults to
**0.5** rather than inheriting 0.8: see the measurement below.
Read-only containment no longer leans on `--max-turns 1`, which was doing
much of it by itself. The deny list gains the agentic and network tools —
`Task` in particular, whose subagent the flag does not reach.
Measured, `skills:eval -- test-servers`:
First move (1 turn) 7/7 at 100%, unchanged
Hand-off (14 turns) 67% and 33% at RUNS=3
So the hand-off is real and unreliable — which is the fact nothing could
observe before this change, and the reason 0.5 rather than 0.8 is the
default bar: at 0.8 both committed cases are red no matter how strongly
the first skill points at the second, and the column stops carrying
signal. A `["pr-flow", "test-servers"]` probe measured 0% and was dropped
rather than kept: `pr-flow` says nothing about fixtures, so the case had
no lever short of broadening a description onto another skill's ground.
Both findings are written up in `docs/skill-authoring.md`.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ABdTnqyBodw6mHVVkGehqe
Signed-off-by: cliffhall <cliff@futurescale.com>
Adds a multi-name filter to `npm run skills:eval` so a set of related
skills can be iterated together, and makes an unknown name a hard error
rather than an empty run — a typo would otherwise report a green 0/0
that reads exactly like a clean pass of the skill that was meant.
Re-aims two `local-dev` cases and one `test-servers` case that partially
overlapped `AGENTS.md`. Partial overlap oscillates across runs rather
than sitting at zero, so it reads as eval noise instead of a defect in
the case; `docs/skill-authoring.md` now records that failure mode and
the observed scores.
Widens the `test-servers` description to the "Use when …" shape and
points the `testing` skill at it, since anything needing a real server
to run against is a procedure `testing` does not carry. That growth
moves the skill listing from 2,921 to 3,144 of 4,000 characters, so the
figure recorded in AGENTS.md is updated to match.
Signed-off-by: cliffhall <cliff@futurescale.com>
`skills:eval` was at 38/46 with several skills measuring 0-33%. The low
scores were mostly bad eval cases rather than bad descriptions, for two
reasons that were nowhere written down:
- The harness runs each case with `--max-turns 1`, so the skill has to be
the model's FIRST tool call. A prompt asking "how does X work here", or
naming a concrete file, sends the model to Read/Grep first — a better
answer to the question, scored as a miss. Every 0% case was one of these.
- `AGENTS.md` is resident on every turn, so a prompt it already answers is
answered without any skill and the case can never pass.
Replace the cases that measured the wrong thing, expand every model-invoked
skill from 3 to 5 positives (negatives kept) so one flake cannot cross the
threshold, and reshape `local-dev`'s description to the action-first form.
Add `docs/skill-authoring.md` with the case shapes that fire, the ones that
cannot, and the probe-then-measure loop.
RUNS=5 npm run skills:eval: 63/63 at or above threshold, 60 at 100%,
all 18 negatives clean.
Closes#2202
Signed-off-by: cliffhall <cliff@futurescale.com>
`disable-model-invocation: true` does not merely make a skill prefer `/name` —
the model cannot invoke it at all, and its description is absent from the skill
listing. So "create a PR for #2163", which is how that work actually starts,
could never reach `pr-flow`. The original rationale ("procedures with side
effects we'd invoke by name anyway") conflates invoking a skill, which only
loads instructions, with performing the procedure it describes.
Flipped `pr-flow`, `pre-push-gate`, `board-ops` and `test-servers` to
model-invocable. `user-invocable` is an independent axis, so this is purely
additive — every `/name` still works. `issue-create`, `issue-triage` and
`release` stay deliberate.
`board-ops` is in the set because `pr-flow` and `issue-create` delegate their
card steps to it: a pointer from a model-invocable skill to a manual-only one is
a dead end for the model.
Costs weighed: the listing share goes 1,199 → 2,256 of 4,000, and a spurious
load costs ~250 characters where a missed one costs a wrong base branch or an
unsigned commit. The rules stay resident in AGENTS.md either way — this improves
the odds of getting the procedure too, it does not become the guarantee.
Eval cases committed for all four, measured at RUNS=3: pre-push-gate, board-ops
and test-servers 5/5; pr-flow 4/5, with "take it through to a pull request" at
100% and the Copilot-review case borderline. Three initial cases were replaced
rather than tuned green: their answers are resident in AGENTS.md, so the model
answering without the skill is correct behaviour, not a miss — the rule this
file already records about what makes a valid trigger case.
AGENTS.md's maintenance rule now says to default to `false` and explains why,
so the advice this change walks back does not get re-applied.
Signed-off-by: cliffhall <cliff@futurescale.com>
- skill-manifest: the closing `---` fence must occupy a whole line. `---oops`
was accepted as a terminator, silently truncating the frontmatter and passing
validation — the exact failure the guard exists to catch. Regression cases
added (prefix, trailing whitespace, EOF, CRLF).
- skill-eval: reject a nonzero `claude -p` exit instead of resolving it as an
empty observation. An auth error or rate limit was scoring as "no skill
invoked", which passes every negative case and reads as a trigger miss on
every positive one.
- skill-eval: extract `collectSkillInvocations` / `sampleHit` behind a `main()`
guard and cover them in `skill-eval.test.mjs` — the stream parsing and the
process-outcome handling are unreachable from an eval run's happy path.
- docs/test-servers: annotate the era on every catalogue row. Six legacy rows
carried no annotation under a "unless noted, Modern" default, so the doc sent
readers to the wrong era for the feature they were trying to see.
- docs/quality-gate: `validate` runs four guards, not three; add the
`verify:skills` row and the two new `test:scripts` suites; fix the `#publishing`
fragment left dangling by the split.
- issue-create / pr-flow: delegate the card steps to `/board-ops` rather than
restating its option IDs, which are explicitly unstable.
- issue-triage: the audit counted presence, not cardinality — an issue carrying
both `v1` and `v2` passed a check that claims "exactly one". Same for the five
type labels. Also scope the two milestone checks to board #28.
- AGENTS.md: codify the v1 milestone exception. Every milestone is a v2 release
bucket, so a v1 issue has none to take and the Incoming-milestone invariant
cannot apply to board #11.
- clients/launcher/README: its pointer promised an explanation of why the cli/tui
manifests ship, which the extracted guide does not make.
Signed-off-by: cliffhall <cliff@futurescale.com>
Closes#2163
`AGENTS.md` was one 1,223-line contract loaded in full on every turn, whether
the task was a one-line CSS fix or a release cut. This splits it: rules stay
resident, procedures move to `.claude/skills/`.
- Ten skills under `.claude/skills/`, seven `disable-model-invocation: true`
(invoked by name only, so they cost no listing context and have no trigger
question) and three model-invoked with committed eval cases.
- `AGENTS.md` 1,223 → 380 lines: rules, a skills index, and a project tree
reduced to one line per entry. The removed recipes are not duplicated.
- `README.md` 811 → 105 lines, split into `docs/{architecture,quality-gate,
test-servers,publishing,docker}.md`, and dropped from `CLAUDE.md`'s auto-load.
- `.github/copilot-instructions.md` and its mirroring rule removed — Copilot
reads `AGENTS.md`.
- `npm run verify:skills` (in `validate`, so in `local:gate` and CI) parses each
SKILL.md the way Claude Code does, requires an explicit invocation mode and
eval cases for model-invoked skills, and budgets the listing.
- `npm run skills:eval` runs the committed trigger cases headless.
Signed-off-by: cliffhall <cliff@futurescale.com>
Cliff Hall