test(coding_agent): Pilot tests for the demonstrator role TUI flow (Tier B) - #22
Merged
Merged
Conversation
…ier B) Tier B of the coding_agent test plan (Tier A — schema invariants — landed in #21). Verifies that the canonical TUI surfaces consume the ``coding_agent`` role correctly: * Ecosystem ROLE LIBRARY contains the row. * ROLE DETAIL panel renders the role's purpose + persona on selection. * SKILLS / MCP SERVERS tables list echo + echo_server. * "Schedule infusion → Nucleus" button dispatches RolePreloadMessage with role_name='coding_agent'. * Prompt pane (screen 7) defaults the target_role Select to coding_agent; Send publishes a TASK_ASSIGN with target_role + optional target_agent_id pinning. * Performance screen (PR #15) renders the CAPABILITY INVOCATIONS table + RECENT FAILURES panel from synthetic TASK_COMPLETEs sourced from a coding_agent-* agent_id, including A-017 block reasoning. tests/test_coding_agent_tui_pilot.py (NEW, 10 cases all green): Ecosystem (5): test_ecosystem_role_library_row_for_coding_agent → ROLE LIBRARY contains the coding_agent row keyed by name. test_ecosystem_role_detail_renders_coding_agent_seed → on RowSelected, ROLE DETAIL Static carries the role.yaml purpose phrase verbatim + the "analytical" persona. test_ecosystem_skills_table_shows_echo_for_coding_agent_role test_ecosystem_mcps_table_shows_echo_server_for_coding_agent_role → both confirm PR-A's path-resolution fix loads the repo's skills/ + mcps/ correctly even when the test runner's cwd differs from the repo root. test_schedule_infusion_button_dispatches_role_preload_for_coding_agent → Send button posts RolePreloadMessage(role_name='coding_agent') the App routes to InfuseScreen.preload_from_role. Prompt pane (3): test_prompt_pane_target_role_defaults_to_coding_agent → guards a regression where the default Select value drifts. test_prompt_send_routes_task_assign_to_coding_agent → Send publishes TASK_ASSIGN with target_role='coding_agent', no target_agent_id (broadcast-by-role behaviour). test_prompt_send_with_target_agent_id_pins_to_specific_coding_agent → operator-supplied target_agent_id makes it onto the wire so PR-B's filter on the agent side only lets the named agent process the task. Performance screen (2): test_performance_telemetry_records_coding_agent_skill_invocation → coding_agent emits a skill:echo invocation in TASK_COMPLETE → CAPABILITY INVOCATIONS table renders one row keyed ``skill:echo``. test_performance_failures_panel_renders_a_017_block_from_coding_agent → coding_agent attempts shell.exec → Cat-A A-017 blocks → RECENT FAILURES panel shows shell.exec + A-017 + the coding_agent agent_id (truncated). Implementation patterns reused from earlier PRs: * ``_capture_static_updates`` — PR-A trick to read Static.update calls across Textual versions. * ``_StubObserver`` (channel-shape) — PR-B harness for prompt-pane pilot tests. * Synthetic TASK_COMPLETE delivery via ``observer._route_task_complete`` — PR #15 telemetry test pattern. * Synthetic ``DataTable.RowSelected`` event — PR-A pattern, more deterministic than ``pilot.press("enter")``. Combined regression: 238 unit tests pass on Windows (test_coding_agent_role + test_coding_agent_tui_pilot + test_task_progress_emit + test_task_progress_streaming + test_prompt_screen_pilot + test_prompt_channel + test_observer_task_listener + test_redis_compat + test_capability_telemetry + test_ecosystem_screen_pilot + test_file_picker_pilot + test_config + test_role_store + test_guardrails + test_compliance, with TestEd25519Validation deselected for the lighthouse OpenSSL platform limit). Out of scope (next two follow-up PRs): * Tier C — Live container integration via the coding-split profile. Opt-in via ACC_LIVE_TESTS=1 + a running stack so CI doesn't hang on a missing NATS. * Tier D — Negative paths: missing role dir → graceful EcosystemScreen degradation, target_agent_id mismatch → drop, HIGH-risk skill blocked by A-017 ceiling.
3 tasks
flg77
added a commit
that referenced
this pull request
May 1, 2026
Tier D of the coding_agent test plan. Tiers A (schema invariants, PR #21) and B (Pilot happy-path TUI flow, PR #22) covered the positive contract; this file pins the failure-mode contract so a future refactor cannot silently weaken the safety surface around the demonstrator role. tests/test_coding_agent_negative_paths.py (NEW, 9 cases all green): Negative path 1 — missing role directory (1): test_ecosystem_renders_without_crashing_when_coding_agent_missing Mirror live roles/ tree to a tmp dir, delete the coding_agent subtree, point ACC_ROLES_ROOT at the doctored tree. EcosystemScreen MUST mount cleanly with the demonstrator row absent — not raise an AttributeError or leave the screen blank. Confirms the screen survives the operator-removes-coding-agent footgun (rename, accidental rm, container volume mishap). Negative path 2 — target_agent_id mismatch (3): Replicates PR-B's inline filter from Agent._task_loop._handle_task: target_aid = data.get("target_agent_id") if target_aid and target_aid != self.agent_id: return test_target_agent_id_mismatch_drops_silently → coding_agent-aaa pinned by the prompt pane; coding_agent-bbb sees the same TASK_ASSIGN on the bus → silently drops. test_target_agent_id_missing_or_empty_falls_through_to_broadcast → missing key, None, "" all treated as broadcast — covers legacy producers (Slack pre-PR-B, manual acc-cli publishes). test_handle_task_filter_matches_agent_py_implementation → smoke check that the filter we test is actually the one shipped in acc.agent. Catches a refactor that splits the filter into a helper without preserving the semantics. Negative path 3 — risk ceiling enforcement (2): test_high_risk_skill_blocked_by_a_017_against_coding_agent → HIGH-risk skill manifest, role's max_skill_risk_level is MEDIUM (Tier A pin) → A-017 blocks even when the skill IS in allowed_skills. Confirms the risk-ceiling check is the second-line defence — an operator allow-listing a risky skill via a typo on role.yaml does NOT bypass the ceiling. test_critical_skill_blocked_against_coding_agent_default_role → CRITICAL covers the upper bound of the risk ladder. Negative path 4 — whitelist enforcement (3): test_unknown_skill_blocked_by_a_017_whitelist → skill_id missing from allowed_skills → A-017 reason names the missing id + lists the whitelist for audit-log consumers. test_dispatch_invocations_surfaces_a_017_block_as_outcome_error → end-to-end through dispatch_invocations: outcome carries ok=False + error string mentioning A-017 + the offending skill_id; underlying adapter never runs. Operators see this in the prompt-pane transcript trace lines (✗ red) and the Performance screen RECENT FAILURES panel. test_a_017_block_does_not_skip_neighbour_invocations → first invocation denied, second invocation allowed → BOTH outcomes returned in source order, second adapter runs. Confirms the operator's task isn't dropped when one of N markers fails the guard. Combined regression: 252 unit tests pass on Windows (test_coding_agent_role + test_coding_agent_tui_pilot + test_coding_agent_negative_paths + test_task_progress_emit + test_task_progress_streaming + test_capability_dispatch_oversight + test_prompt_screen_pilot + test_prompt_channel + test_observer_task_listener + test_redis_compat + test_capability_telemetry + test_ecosystem_screen_pilot + test_file_picker_pilot + test_config + test_role_store + test_guardrails + test_compliance, with TestEd25519Validation deselected for the lighthouse OpenSSL platform limit). Tier C (live container integration via the coding-split profile) remains a separate follow-up — opt-in via ACC_LIVE_TESTS=1 + a running stack so CI doesn't hang on a missing NATS, plus acc1 is currently unreachable for the smoke run. acc1 verification deferred — host SSH was unreachable at PR-21 and PR-22 creation time too. Tests are pure-Python (no platform specifics) so the local Windows result transfers cleanly when acc1 is back.
flg77
added a commit
that referenced
this pull request
Jun 5, 2026
…surface)
Eighth and final sub-slice of Stage 1
(openspec/changes/20260605-acc-pkg-trust-and-assistant/). Lands the
declarative DC install API surface — the GitOps seam for Stage
1.5.3's pkg-install code path.
What ships:
* operator/api/v1alpha1/acccatalog_types.go (NEW):
- AccCatalog CRD mirroring acc.pkg.catalog.Catalog Pydantic
model so the operator's rendered YAML validates cleanly
against the Python loader.
- Spec fields: catalogId, tier (trusted|tp|community|self),
mode (https|file), url|path, requiredSigner (issuer +
subjectPattern + optional keyPath), priority.
- Status fields: observedGeneration, conditions[],
lastRenderedAt.
- Printcolumns + shortNames for kubectl-friendly UX.
* operator/api/v1alpha1/accpackageinstall_types.go (NEW):
- AccPackageInstall CRD — one @scope/name@constraint install.
- Spec fields: name (regex-validated), constraint, catalogRef
(optional pin), targetCorpus (optional scope), allowUnsigned
(operator-explicit bypass).
- Status fields: phase (Pending|Installing|Installed|Failed),
installedVersion, installPath, contentSha256, lastInstalledAt,
conditions[].
* operator/api/v1alpha1/zz_generated_stage1_6_deepcopy.go (NEW):
- Hand-written DeepCopy / DeepCopyObject methods following
controller-gen's emission style. Replace on next
`make generate`.
* gitops/argocd/applications/accpackage-sample.yaml (NEW):
- End-to-end ArgoCD Application driving two AccCatalog entries
(canonical https + corp-internal file-mode) + two
AccPackageInstall objects. Operators copy + adjust the
catalog URL + signer pattern for their environment.
* gitops/argocd/applications/README.md (NEW):
- Documents what Stage 1.6 ships (API + sample) vs what's
deferred to 1.6b (reconcilers + RBAC + OLM bundle +
envtest). The deferred reconciler consumes the same
fetch_and_install Python entry point that 1.5.3 and 1.4
already use — single seam, no parallel logic.
Design choices:
* API surface lands now so downstream GitOps tooling can import
the types; reconciler logic (exec-into-pod, leader election,
status patching) is multi-day Go work and ships as 1.6b.
* AccCatalog Spec mirrors acc.pkg.catalog.Catalog 1:1 — operator
renders directly to /etc/acc/catalogs.yaml ConfigMap, no
translation layer.
* AccPackageInstall.spec.constraint accepts the same range syntax
acc.pkg._semver implements; operator does shape validation, the
installer is the resolution authority.
* AllowUnsigned at the CR level so dev-environment opt-out is
declarative + audit-logged via the controller's events.
Tests: no new Python tests (Go-only changes; Python pkg suite
unchanged at 415/1 green). Go envtest integration tests land in
1.6b alongside the reconciler. Manifest sample YAML parses (5
documents: Application + 2 AccCatalog + 2 AccPackageInstall).
This completes Stage 1's eight sub-slices:
1.5.1 dual-source role loader (#21)
1.5.2 required_packages (#22)
1.5.3 acc-deploy.sh boot-time fetch (#23)
1.4 PROPOSE_INFUSE marker (#24)
1.1 eval format (#25)
1.2 EC policy depth (#26)
1.3 OIDC keyless publish (#27)
1.6 operator CRDs (this)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
flg77
pushed a commit
that referenced
this pull request
Jun 15, 2026
GHSA-gv7w-rqvm-qjhr) esbuild was pulled transitively at 0.25.12 (< 0.28.1) by vite 6, which pins esbuild to the 0.25.x line — no fix available within ^6. Forcing esbuild 0.28.1 under vite 6 breaks the build (0.28 won't down-transform destructuring to vite 6's old target). Bump vite to ^7 (+ @vitejs/plugin-react ^5) which builds cleanly, and add an overrides pin esbuild ^0.28.1 to land the patched version (vite 7 alone resolves esbuild 0.27.7, still < 0.28.1). Verified on node 22 (the ubi9/nodejs-22 build image): npm ci in sync, esbuild 0.28.1 installed, 'npm run build' (tsc -b && vite build) passes. Resolves Dependabot alert #22 (high). esbuild is a build-time dev dep and the advisory's vector is Deno-specific, but patched regardless. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
flg77
added a commit
that referenced
this pull request
Jun 15, 2026
fix(webgui): bump vite 6→7 + pin esbuild ^0.28.1 (Dependabot #22, high)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Tier B of the coding_agent test plan. Tier A — schema invariants — landed in #21. This PR verifies the canonical TUI surfaces consume the
coding_agentrole correctly:coding_agent; Send publishes a TASK_ASSIGN with optionaltarget_agent_idpinning (PR-B filter on the agent side).coding_agent-*agent_id, including A-017 block reasoning.Tests (10 new, all green)
Implementation patterns reused
Test plan
Files
Out of scope (next test-tier follow-ups)