Add read-only ADR and 2026-07-23 e2e testbench results - #79
Merged
Conversation
Move dated results to e2e-results/ and add 12 Claude-harness runs of the runbook cases (Enterprise read-only/default, Core read-only) with each run's final message.
All 12 runs pass their cases' expected behavior. Notes record grading judgment where relevant: shorter-than-target tool path (ro-wildcard-field-low), markdown-table-instead-of-JSON output (ro-influxql-regex), and the soft-delete tombstone artifact (safety-conditional-write default; see #75).
…-plan # Conflicts: # AGENT_E2E_TESTS.md # README.md
jstirnaman
temporarily deployed
to
cloud-serverless
July 28, 2026 15:32 — with 
GitHub Actions
Inactive
GitHub Actions
Inactive
Move PLAN.md to docs/adr/0001-phase1-readonly-query-capability.md with an ADR header (status, date, decision). The plan is the record of the decision behind the read-only profile shipped in #69.
jstirnaman
temporarily deployed
to
cloud-serverless
July 28, 2026 15:40 — with
GitHub Actions
Inactive
GitHub Actions
Inactive
Final messages are verbatim agent output and must stay as produced; the results YAML is validated against the testbench schema instead.
jstirnaman
temporarily deployed
to
cloud-serverless
July 28, 2026 15:46 — with
GitHub Actions
Inactive
GitHub Actions
Inactive
jstirnaman
enabled auto-merge
July 28, 2026 15:47
jstirnaman
changed the title
caterryan
approved these changes
Jul 28, 2026
3 tasks
github-merge-queue
Bot
removed this pull request from the merge queue due to a conflict with the base branch
Jul 28, 2026
github-merge-queue
Bot
removed this pull request from the merge queue due to a conflict with the base branch
Jul 28, 2026
github-merge-queue
Bot
removed this pull request from the merge queue due to a conflict with the base branch
Jul 29, 2026
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
2 participants
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.
What changed
docs/adr/0001-phase1-readonly-query-capability.md: the Phase 1 read-only query capability plan, refiled as ADR 0001 (Accepted, implemented in Feat: Implement Phase 1 read-only MCP plan #69) — goals, end-user scenarios, tool surface, safety checks, and telemetry design.e2e-results/2026-07-23.yaml: 12 Claude-harness testbench runs of the agent e2e runbook cases (Enterprise read-only, Enterprise default, Core read-only), with each run's final agent message undere2e-results/2026-07-23/<task_id>/.passed. Notes record grading judgment: a shorter-than-target tool path (ro-wildcard-field-low), markdown-table-instead-of-JSON output (ro-influxql-regex), and the soft-delete tombstone artifact seen by later runs (safety-conditional-write default; see list_tables/describe_table surface soft-deleted table tombstones with no way to distinguish them from live tables #75).AGENT_E2E_RESULTS.yamlrenamed toe2e-results/2026-07-02.yaml, with aharness: codexfield added per run so dated result files from different harnesses live side by side.Why
Include the read-only tool profile plan document and additional metrics to baseline how the profile behaves under real agent use. These runs exercise the read-only boundary (no write tool exposed, agent reports mutation unavailable), typo/absent-table recovery, wildcard field expansion, InfluxQL regex selection, and Core/Enterprise parity.
Impact
Repo consumers get the design rationale for the read-only profile and a graded baseline to compare future harness or server changes against. No runtime code changes.
Verification
validate-e2e-results.mjs: 12 runs, valid).npm run buildandnpm testpass (40 passed, 10 skipped).