A catalog entry may now carry its own `maxTokens`, matching the shape the
pi-ai adapter already exposes. Exact-model resolution prefers it over the
profile value, so capping one model no longer means capping the route.
Defaults are unchanged: an entry without a cap, and any unlisted
pass-through id, still resolve to the profile `maxTokens` (256,000), and the
shipped catalog keeps its context windows.
Five findings from the #939 review, each reproduced before being fixed.
**Configuration reads are as privileged as writes.** `settings.describe`
returns every exposed namespace's configuration and `credentials.describe`
reports whether an arbitrary environment-variable name is configured and from
where — reconnaissance no anonymous caller should have. Both join
PRIVILEGED_METHODS, so the whole configuration plane is loopback-only until
real authentication exists; `trustedHosts` was never authentication. The model
catalog stays reachable: it carries no endpoints or key state, and a LAN
client's model picker legitimately needs it. Asserted over a real HTTP server,
because the Host header a browser actually sends is what decides this.
**The proxy serves only namespaces a registered model provider addresses.**
The settings seam is general — any plugin may register one — but the Web
configuration plane is the model-provider surface. Without the gate, every
future `settings.register()` would silently become remotely readable and
writable configuration. An unregistered namespace and an unexposed one answer
identically, so no caller can enumerate the registry one probe at a time.
**Path-addressed writes replace the redacted-document rebuild.** The editor
reads the REDACTED descriptor, so rebuilding a section from it and replacing
wholesale deleted every literal secret the wire never returned — reproduced as
`{baseURL, reasoning}` in, stored `apiKey` gone out. `settings.mutate` applies
set/unset ops to the section as it stands at the front of the seam's write
queue, and the client names only fields it can see, so an unseen secret is
untouched by construction rather than by care.
P2s in the same pass: `llm/adapters-updated` now contains async listener
rejections (an uncontained one escaped as unhandledRejection, contradicting
the documented "observer failures are contained"); llm-deepseek's retry-policy
swap uses the atomic `registration.replace` instead of dispose-then-register,
which published `[]` then `["deepseek-official"]` so an observer saw the
provider disappear and come back; and a transport rejection no longer strands
the page in `loading` or a card in `busy`, with removal failures surfaced on
the page banner instead of swallowed.
The policy home's resolve() now stamps readDenyPaths, so every consumer
that pins the resolved shape (bash-sandbox hand-off, tool-fs stamps)
carries it, and three uncovered branches gained real tests: landlock
reporting partial enforcement for a denial it cannot express, the
policy's default under programmatic construction, and both ambient
credential paths in llm-deepseek without a mounted seam.
Four review findings across the seam and both adapters.
registerAdapter now returns a handle carrying replace(providers): the
candidate route set is validated in full before anything moves, so a
route another adapter owns leaves the previous registration intact, and
the swap itself is one synchronous section with no observable gap. pi-ai
uses it instead of dispose-then-register — the old shape dropped every
route when the new set conflicted, and its facts cache could then equal
the registry's, so reverting to a working configuration never re-applied.
Its registration facts are also sorted by provider, so a settings
document that merely reorders keys no longer triggers a swap.
DeepSeek's per-request snapshot now carries the credential facts, and
resolveApiKey receives it instead of re-reading the raw config: a
settings generation the resolver rejects can no longer contribute its
literal key to a request the previous generation's endpoint serves.
pi-ai only defers to the SDK's provider-native discovery when a profile
names no credential at all; a configured apiKeyEnv that misses now fails
with MISSING_CREDENTIAL naming the route and the reference, instead of
handing pi-ai undefined and letting it authenticate with an unrelated
ambient key.
The eager boot-time credential probe is gone: it could run before the
credentials service mounted and reported every failure as a missing key.
The route stays registered and browsable; the first request gives the
accurate error, whose guidance now leads with the credential store and
mentions a literal apiKey last.
The shipped surfaces loaded the harness home's .env into the process
environment before cordis booted. credentials-local then saw every stored
key as an ambient launch override: describe reported source 'env' with
writable false, and set/unset rejected as shadowed — so a key the web page
or TUI stored was unrotatable and undeletable from the next run onward,
and the adapter kept using the value captured at launch.
The home's .env is now the credential provider's own store, read by that
provider alone and hot-reloaded by it. The genuine launch environment and
the invoking directory's .env (loaded by the bin) remain the read-only
ambient layer, so a plain composition without the provider still resolves
keys exactly as before.
Proven by a real restart in the loader composition: store a key through
the seam, dispose the tree, re-boot over the same harness home, and the
entry is still file-sourced and writable — rotating it lands on the very
next request.
ctx.llm gains 'llm/adapters-updated' — a payload-free registry notification
emitted at every topology commit point (adapter routes registering or
disposing, directory entries appearing or withdrawing) with contained
observers and INVARIANT rethrow — plus registerConfigurableProviders/
listConfigurableProviders, the directory of routes an adapter plugin can
activate through configuration. llm-deepseek declares deepseek-official
(whole llm-deepseek section as profile); llm-pi-ai declares the full
installed catalog under providers.<provider> even while dormant, so the
web settings surface can offer every provider before any route exists.
The invariant companion asserts the registry stays readable at each
notification.
The native adapter's route was named deepseek, colliding with pi-ai's
catalog provider of the same name, so the two DeepSeek paths could never
be mounted side by side. The web settings page needs both configurable at
once. Compositions, fixtures, goldens, scaffolding defaults, and docs all
move together (pre-release, no shim); TUI/session-query-spill/
missing-credential goldens re-recorded through their keyless refresh
modes because provider-name length shifts box padding and spill
truncation points.
llm-deepseek gains a Loader+Include composition spec proving external
settings.yaml/.env edits reach the very next request, and a real-API e2e
where only a credentials-local document holds the key. The headless example
pins the first-run missing-credential UX as a keyless stream-json snapshot
(new credentials.cordis.snapshot.yml scenario); runLoaderSmoke learns
expectedExitCode so a designed failure surface can be pinned instead of
masked.
The adapter takes an options thunk and a per-stream credential resolver
instead of frozen construction facts: base URL, catalog, defaults, idle
budget, and the API key re-resolve at each operation, so a settings or
credential change reaches the very next request while in-flight streams
keep the facts they started with. resolveAdapterOptions is the one explicit
resolve step (entry config fails loud at load; a live snapshot failing a
beyond-schema bound keeps the last good options). The plugin layers its
entry config under the optional llm-deepseek settings section and resolves
keys literal-first through ctx.credentials with an ambient env fallback; a
missing key now registers the route, warns, and fails each request with
actionable MISSING_CREDENTIAL instead of failing plugin load. The
registration-captured retry policy re-registers the route in place when it
changes.
Implements the approved simplification Agent Note: sse.ts now pipes the
response body through TextDecoderStream and EventSourceParserStream
(eventsource-parser/stream) and keeps only the DeepSeek protocol shim —
yield each event's data, terminate on [DONE], throw
LlmError('STREAM_CLOSED') on EOF without the sentinel. The SSE
spec-conformance tests are deleted; sse.spec.ts pins only the
[DONE]/STREAM_CLOSED/EOF contract, including the new spec-strict verdict
that an unterminated trailing event is truncation (the old parser
flushed it — a robustness nicety no real provider shape needs).
eventsource-parser@^3.1.0 becomes llm-deepseek's second runtime
dependency (already in the lockfile transitively via the MCP SDK).
Docs: the Agent Note moves proposed/ → implemented/ and is rewritten per
the lifecycle contract; the rejected NIH roll-up note's inbound links
follow. The twin-adapters note, dsh-llm LlmAdapter JSDoc (and its
type-equiv fences), cookbook, group/package READMEs, root AGENTS.md
layout line, sdk-helper comments, and the regenerated config catalog
drop the "hand-rolled fetch + SSE" claim in both languages; all eight
touched pairs re-recorded.
A well-formed provider stream that ends with finish_reason stop and zero
content blocks previously became a successful empty assistant message: the
turn completed silently, and drivers like goal-session counted the no-op
round. Both adapters now map that degenerate completion to a finish
{kind:'error'} with the new canonical EMPTY_RESPONSE code from dsh-llm, and
dsh-llm-retry adds the code to its default retryable set, so the existing
closed-step recovery path retries it and fails loud once the budget is
exhausted.
Covered by adapter unit tests, an llm-retry default-policy test, and a new
authored keyless ACP snapshot (empty-response-retry) with a deterministic
1 ms zero-jitter retry overlay.
17 master merges (GUI host/web stack, compact header, dshweb, fast local
hooks, windows coverage skips, worktree runtime). Conflicts:
- packages/ui/user-interaction/src/index.ts: master extracted the question
types to types.ts; took the re-export and grafted our detail field (the
plan-review payload) into the extracted AskUserQuestionItem.
- generated catalogs regenerated over both sides.
The module graph crossed mermaid's default 500-edge render guard with the
GUI packages plus our plan-mode edges; raised maxEdges in verify-mermaid
(a secure config settable only at initialize).
Note: the remote branch also renamed packages/mode -> packages/plan
(PR #512 plan-mode-simplification, fast-forwarded before this merge).
Master arrivals re-grafted onto the plan-mode surfaces:
- Package invariant runtime contracts: every package owns ./invariant —
added the dsh-mode companion (mode/set payload validation: non-empty
trimmed name), its unit spec, exports/peer wiring, and the mode group in
the tsconfig invariant paths glob.
- Interception waterfalls gained an AbortSignal parameter
(agent/prompt-submit, agent/turn-continuation) and ToolExecutionInput.signal
became required: listeners and test dispatch updated; the exit tool now
forwards exec.signal unconditionally.
- TUI question dialog: master's position/unanswered header layout kept,
the plan-review detail block re-grafted between question and answers.
- LSP capability family arrived: lsp tool joins the catalog pin, dsh-lsp
examples deps beside dsh-mode; lsp-definition snapshot refreshed with
the mode command and modes advertisement.
- docs/architecture.md and packages/README.md regenerated tables re-grafted
(ctx.modes row, mode package row) and condensed within word ceilings.
- examples/acp-agent cordis.yml: kept the deployment-owned plan instructions
(incl. the conversational-agreement sentence) beside master's routed-model
compaction comment; snapshot expected outputs refreshed keyless and noise
fixtures restored.
A TUI run against an unreachable endpoint failed with only 'fetch failed':
undici wraps transport failures in a bare TypeError whose diagnosis lives
on .cause, and every diagnostic seam rendered only error.message. The
readline front door additionally rendered failed turns as pure silence.
- dsh-llm: new errorChain(value) renders the full cause chain and
AggregateError members with circular/hostile-coercion containment.
- llm-deepseek: pre-response transport failures throw LlmError('NETWORK')
naming the endpoint and chaining the fetch TypeError; aborts keep their
DOMException so the loop still classifies them as cancellation.
- agent-loop: durable turn/end error messages and logger warnings render
through errorChain; local renderThrown copies removed.
- ui-stdio: failure turn/end reasons now render ([turn failed <code>],
[turn aborted], [turn rejected], output-token-limit); startup-failure
logs use errorChain.
- ui-tui: agent/error notices and the startup-failure line use errorChain.
