foretop-telltale 0.1.0__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (48) hide show
  1. foretop_telltale-0.1.0/LICENSE +202 -0
  2. foretop_telltale-0.1.0/PKG-INFO +354 -0
  3. foretop_telltale-0.1.0/README.md +339 -0
  4. foretop_telltale-0.1.0/pyproject.toml +72 -0
  5. foretop_telltale-0.1.0/pyproject.toml.orig +81 -0
  6. foretop_telltale-0.1.0/src/keel/__init__.py +1 -0
  7. foretop_telltale-0.1.0/src/keel/collect/__init__.py +0 -0
  8. foretop_telltale-0.1.0/src/keel/collect/line_tracking.py +48 -0
  9. foretop_telltale-0.1.0/src/keel/collect/walk.py +95 -0
  10. foretop_telltale-0.1.0/src/keel/finding.py +105 -0
  11. foretop_telltale-0.1.0/src/keel/gate.py +152 -0
  12. foretop_telltale-0.1.0/src/keel/git_ref.py +45 -0
  13. foretop_telltale-0.1.0/src/keel/owner.py +58 -0
  14. foretop_telltale-0.1.0/src/keel/render/__init__.py +0 -0
  15. foretop_telltale-0.1.0/src/keel/render/annotations.py +62 -0
  16. foretop_telltale-0.1.0/src/keel/render/markdown.py +40 -0
  17. foretop_telltale-0.1.0/src/keel/render/sarif.py +85 -0
  18. foretop_telltale-0.1.0/src/keel/render/terminal.py +38 -0
  19. foretop_telltale-0.1.0/src/keel/render/triage_json.py +15 -0
  20. foretop_telltale-0.1.0/src/keel/report.py +98 -0
  21. foretop_telltale-0.1.0/src/keel/run.py +67 -0
  22. foretop_telltale-0.1.0/src/keel/triage.py +167 -0
  23. foretop_telltale-0.1.0/src/telltale/__init__.py +1 -0
  24. foretop_telltale-0.1.0/src/telltale/__main__.py +4 -0
  25. foretop_telltale-0.1.0/src/telltale/brand.py +6 -0
  26. foretop_telltale-0.1.0/src/telltale/check.py +121 -0
  27. foretop_telltale-0.1.0/src/telltale/cli.py +174 -0
  28. foretop_telltale-0.1.0/src/telltale/findings.py +238 -0
  29. foretop_telltale-0.1.0/src/telltale/git_ref.py +44 -0
  30. foretop_telltale-0.1.0/src/telltale/grafana.py +106 -0
  31. foretop_telltale-0.1.0/src/telltale/match.py +97 -0
  32. foretop_telltale-0.1.0/src/telltale/models.py +121 -0
  33. foretop_telltale-0.1.0/src/telltale/openapi.py +92 -0
  34. foretop_telltale-0.1.0/src/telltale/otel.py +150 -0
  35. foretop_telltale-0.1.0/src/telltale/prometheus.py +62 -0
  36. foretop_telltale-0.1.0/src/telltale/promql.py +75 -0
  37. foretop_telltale-0.1.0/src/telltale/regression.py +68 -0
  38. foretop_telltale-0.1.0/src/telltale/render/__init__.py +0 -0
  39. foretop_telltale-0.1.0/src/telltale/render/annotations.py +48 -0
  40. foretop_telltale-0.1.0/src/telltale/render/json_renderer.py +15 -0
  41. foretop_telltale-0.1.0/src/telltale/render/markdown.py +96 -0
  42. foretop_telltale-0.1.0/src/telltale/render/sarif.py +12 -0
  43. foretop_telltale-0.1.0/src/telltale/render/terminal.py +119 -0
  44. foretop_telltale-0.1.0/src/telltale/response_code.py +22 -0
  45. foretop_telltale-0.1.0/src/telltale/route_template.py +14 -0
  46. foretop_telltale-0.1.0/src/telltale/rules/label_names.yaml +58 -0
  47. foretop_telltale-0.1.0/src/telltale/triage.py +150 -0
  48. foretop_telltale-0.1.0/src/telltale/walk.py +26 -0
@@ -0,0 +1,202 @@
1
+
2
+ Apache License
3
+ Version 2.0, January 2004
4
+ http://www.apache.org/licenses/
5
+
6
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
7
+
8
+ 1. Definitions.
9
+
10
+ "License" shall mean the terms and conditions for use, reproduction,
11
+ and distribution as defined by Sections 1 through 9 of this document.
12
+
13
+ "Licensor" shall mean the copyright owner or entity authorized by
14
+ the copyright owner that is granting the License.
15
+
16
+ "Legal Entity" shall mean the union of the acting entity and all
17
+ other entities that control, are controlled by, or are under common
18
+ control with that entity. For the purposes of this definition,
19
+ "control" means (i) the power, direct or indirect, to cause the
20
+ direction or management of such entity, whether by contract or
21
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
22
+ outstanding shares, or (iii) beneficial ownership of such entity.
23
+
24
+ "You" (or "Your") shall mean an individual or Legal Entity
25
+ exercising permissions granted by this License.
26
+
27
+ "Source" form shall mean the preferred form for making modifications,
28
+ including but not limited to software source code, documentation
29
+ source, and configuration files.
30
+
31
+ "Object" form shall mean any form resulting from mechanical
32
+ transformation or translation of a Source form, including but
33
+ not limited to compiled object code, generated documentation,
34
+ and conversions to other media types.
35
+
36
+ "Work" shall mean the work of authorship, whether in Source or
37
+ Object form, made available under the License, as indicated by a
38
+ copyright notice that is included in or attached to the work
39
+ (an example is provided in the Appendix below).
40
+
41
+ "Derivative Works" shall mean any work, whether in Source or Object
42
+ form, that is based on (or derived from) the Work and for which the
43
+ editorial revisions, annotations, elaborations, or other modifications
44
+ represent, as a whole, an original work of authorship. For the purposes
45
+ of this License, Derivative Works shall not include works that remain
46
+ separable from, or merely link (or bind by name) to the interfaces of,
47
+ the Work and Derivative Works thereof.
48
+
49
+ "Contribution" shall mean any work of authorship, including
50
+ the original version of the Work and any modifications or additions
51
+ to that Work or Derivative Works thereof, that is intentionally
52
+ submitted to Licensor for inclusion in the Work by the copyright owner
53
+ or by an individual or Legal Entity authorized to submit on behalf of
54
+ the copyright owner. For the purposes of this definition, "submitted"
55
+ means any form of electronic, verbal, or written communication sent
56
+ to the Licensor or its representatives, including but not limited to
57
+ communication on electronic mailing lists, source code control systems,
58
+ and issue tracking systems that are managed by, or on behalf of, the
59
+ Licensor for the purpose of discussing and improving the Work, but
60
+ excluding communication that is conspicuously marked or otherwise
61
+ designated in writing by the copyright owner as "Not a Contribution."
62
+
63
+ "Contributor" shall mean Licensor and any individual or Legal Entity
64
+ on behalf of whom a Contribution has been received by Licensor and
65
+ subsequently incorporated within the Work.
66
+
67
+ 2. Grant of Copyright License. Subject to the terms and conditions of
68
+ this License, each Contributor hereby grants to You a perpetual,
69
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
70
+ copyright license to reproduce, prepare Derivative Works of,
71
+ publicly display, publicly perform, sublicense, and distribute the
72
+ Work and such Derivative Works in Source or Object form.
73
+
74
+ 3. Grant of Patent License. Subject to the terms and conditions of
75
+ this License, each Contributor hereby grants to You a perpetual,
76
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
77
+ (except as stated in this section) patent license to make, have made,
78
+ use, offer to sell, sell, import, and otherwise transfer the Work,
79
+ where such license applies only to those patent claims licensable
80
+ by such Contributor that are necessarily infringed by their
81
+ Contribution(s) alone or by combination of their Contribution(s)
82
+ with the Work to which such Contribution(s) was submitted. If You
83
+ institute patent litigation against any entity (including a
84
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
85
+ or a Contribution incorporated within the Work constitutes direct
86
+ or contributory patent infringement, then any patent licenses
87
+ granted to You under this License for that Work shall terminate
88
+ as of the date such litigation is filed.
89
+
90
+ 4. Redistribution. You may reproduce and distribute copies of the
91
+ Work or Derivative Works thereof in any medium, with or without
92
+ modifications, and in Source or Object form, provided that You
93
+ meet the following conditions:
94
+
95
+ (a) You must give any other recipients of the Work or
96
+ Derivative Works a copy of this License; and
97
+
98
+ (b) You must cause any modified files to carry prominent notices
99
+ stating that You changed the files; and
100
+
101
+ (c) You must retain, in the Source form of any Derivative Works
102
+ that You distribute, all copyright, patent, trademark, and
103
+ attribution notices from the Source form of the Work,
104
+ excluding those notices that do not pertain to any part of
105
+ the Derivative Works; and
106
+
107
+ (d) If the Work includes a "NOTICE" text file as part of its
108
+ distribution, then any Derivative Works that You distribute must
109
+ include a readable copy of the attribution notices contained
110
+ within such NOTICE file, excluding those notices that do not
111
+ pertain to any part of the Derivative Works, in at least one
112
+ of the following places: within a NOTICE text file distributed
113
+ as part of the Derivative Works; within the Source form or
114
+ documentation, if provided along with the Derivative Works; or,
115
+ within a display generated by the Derivative Works, if and
116
+ wherever such third-party notices normally appear. The contents
117
+ of the NOTICE file are for informational purposes only and
118
+ do not modify the License. You may add Your own attribution
119
+ notices within Derivative Works that You distribute, alongside
120
+ or as an addendum to the NOTICE text from the Work, provided
121
+ that such additional attribution notices cannot be construed
122
+ as modifying the License.
123
+
124
+ You may add Your own copyright statement to Your modifications and
125
+ may provide additional or different license terms and conditions
126
+ for use, reproduction, or distribution of Your modifications, or
127
+ for any such Derivative Works as a whole, provided Your use,
128
+ reproduction, and distribution of the Work otherwise complies with
129
+ the conditions stated in this License.
130
+
131
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
132
+ any Contribution intentionally submitted for inclusion in the Work
133
+ by You to the Licensor shall be under the terms and conditions of
134
+ this License, without any additional terms or conditions.
135
+ Notwithstanding the above, nothing herein shall supersede or modify
136
+ the terms of any separate license agreement you may have executed
137
+ with Licensor regarding such Contributions.
138
+
139
+ 6. Trademarks. This License does not grant permission to use the trade
140
+ names, trademarks, service marks, or product names of the Licensor,
141
+ except as required for reasonable and customary use in describing the
142
+ origin of the Work and reproducing the content of the NOTICE file.
143
+
144
+ 7. Disclaimer of Warranty. Unless required by applicable law or
145
+ agreed to in writing, Licensor provides the Work (and each
146
+ Contributor provides its Contributions) on an "AS IS" BASIS,
147
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
148
+ implied, including, without limitation, any warranties or conditions
149
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
150
+ PARTICULAR PURPOSE. You are solely responsible for determining the
151
+ appropriateness of using or redistributing the Work and assume any
152
+ risks associated with Your exercise of permissions under this License.
153
+
154
+ 8. Limitation of Liability. In no event and under no legal theory,
155
+ whether in tort (including negligence), contract, or otherwise,
156
+ unless required by applicable law (such as deliberate and grossly
157
+ negligent acts) or agreed to in writing, shall any Contributor be
158
+ liable to You for damages, including any direct, indirect, special,
159
+ incidental, or consequential damages of any character arising as a
160
+ result of this License or out of the use or inability to use the
161
+ Work (including but not limited to damages for loss of goodwill,
162
+ work stoppage, computer failure or malfunction, or any and all
163
+ other commercial damages or losses), even if such Contributor
164
+ has been advised of the possibility of such damages.
165
+
166
+ 9. Accepting Warranty or Additional Liability. While redistributing
167
+ the Work or Derivative Works thereof, You may choose to offer,
168
+ and charge a fee for, acceptance of support, warranty, indemnity,
169
+ or other liability obligations and/or rights consistent with this
170
+ License. However, in accepting such obligations, You may act only
171
+ on Your own behalf and on Your sole responsibility, not on behalf
172
+ of any other Contributor, and only if You agree to indemnify,
173
+ defend, and hold each Contributor harmless for any liability
174
+ incurred by, or claims asserted against, such Contributor by reason
175
+ of your accepting any such warranty or additional liability.
176
+
177
+ END OF TERMS AND CONDITIONS
178
+
179
+ APPENDIX: How to apply the Apache License to your work.
180
+
181
+ To apply the Apache License to your work, attach the following
182
+ boilerplate notice, with the fields enclosed by brackets "[]"
183
+ replaced with your own identifying information. (Don't include
184
+ the brackets!) The text should be enclosed in the appropriate
185
+ comment syntax for the file format. We also recommend that a
186
+ file or class name and description of purpose be included on the
187
+ same "printed page" as the copyright notice for easier
188
+ identification within third-party archives.
189
+
190
+ Copyright 2026 the foretop authors
191
+
192
+ Licensed under the Apache License, Version 2.0 (the "License");
193
+ you may not use this file except in compliance with the License.
194
+ You may obtain a copy of the License at
195
+
196
+ http://www.apache.org/licenses/LICENSE-2.0
197
+
198
+ Unless required by applicable law or agreed to in writing, software
199
+ distributed under the License is distributed on an "AS IS" BASIS,
200
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
201
+ See the License for the specific language governing permissions and
202
+ limitations under the License.
@@ -0,0 +1,354 @@
1
+ Metadata-Version: 2.4
2
+ Name: foretop-telltale
3
+ Version: 0.1.0
4
+ Summary: Compares an OpenAPI contract with the telemetry a service actually emits, and produces a route-by-route gap list before an incident finds it for you.
5
+ License-Expression: Apache-2.0
6
+ License-File: LICENSE
7
+ Requires-Dist: typer>=0.15
8
+ Requires-Dist: rich>=13.9
9
+ Requires-Dist: pathspec>=0.12
10
+ Requires-Dist: pyyaml>=6.0
11
+ Requires-Dist: pydantic>=2.9
12
+ Requires-Dist: httpx>=0.28
13
+ Requires-Python: >=3.12
14
+ Description-Content-Type: text/markdown
15
+
16
+ # telltale
17
+
18
+ > Compares an OpenAPI contract with the telemetry a service actually emits, and produces a
19
+ > route-by-route gap list before an incident finds it for you.
20
+
21
+ An OpenAPI spec says a route exists. Somewhere a Prometheus alert, an OTel Collector pipeline,
22
+ or a Grafana dashboard is supposed to be watching it — or isn't, and nobody notices until the
23
+ incident. `telltale` compares the two and tells you which routes (and which of their
24
+ documented 4xx/5xx failures specifically) have no telemetry at all.
25
+
26
+ ## Install and run
27
+
28
+ ```
29
+ uvx foretop-telltale check .
30
+ ```
31
+
32
+ No config, no account, no network call beyond installing the package — autodetects an OpenAPI
33
+ document at the repo root and walks the repo for Prometheus alert rules, an OTel Collector
34
+ config, and Grafana dashboards, entirely local. Add `--base origin/main` to fail only when
35
+ coverage that existed at the merge base is gone now (never an absolute threshold — see DEC-04
36
+ below), or `--format markdown`/`--format annotations` for the same output a PR comment or
37
+ GitHub annotation would use. `telltale check --help` lists everything.
38
+
39
+ ## Use it as a GitHub Action
40
+
41
+ ```yaml
42
+ - uses: foretop-dev/telltale@main
43
+ with:
44
+ base: ${{ github.event.pull_request.base.sha }} # default; empty string disables the gate
45
+ ```
46
+
47
+ Annotates the pull request inline and upserts one summary comment (finds its own previous
48
+ comment by an HTML marker and edits it — never posts a second one on repeat pushes). Needs
49
+ `permissions: pull-requests: write` in the calling workflow.
50
+
51
+ ## License
52
+
53
+ Apache-2.0. This file tracks build status below; the source-of-truth product spec lives in the
54
+ private monorepo this is mirrored from (see `scripts/sync-mirror.sh` there).
55
+
56
+ ## Status
57
+
58
+ **Slice 1** (Session 9): OpenAPI parser + Prometheus alerting-rule parser + the matcher +
59
+ a terminal coverage matrix, run against one fixture service. No network, no DB.
60
+
61
+ **Slice 2, part 1** (Session 11): OTel Collector config parser + Grafana dashboard parser,
62
+ extending real coverage to all four signal dimensions (`log`/`metric`/`trace`/`alert`) —
63
+ every column in the matrix now has at least one real extractor behind it, none are
64
+ permanently `ambiguous` by construction anymore (see below for what `ambiguous` still means).
65
+
66
+ **Slice 3 (Session 12): the error-mode dimension** — DEC-03's own words, "the actual
67
+ differentiator, do not skip it." Every route-level cell now has siblings, one per documented
68
+ 4xx/5xx response code (`src/telltale/response_code.py` normalizes both OpenAPI's "5XX"
69
+ wildcard and specific codes like "404" into one comparable space) — the coverage question
70
+ becomes not just "does this route have an alert" but "does *this specific documented
71
+ failure* have one." `telltale check` now renders two tables: route-level, then error-mode
72
+ (only when at least one route documents a 4xx/5xx code).
73
+
74
+ **Slice 4, part 1 (Session 13): the regression-vs-merge-base gate** — DEC-04: "Fail on
75
+ regression... vs the merge base. Absolute thresholds get disabled on day two of adoption." No
76
+ `--fail-on <threshold>` flag exists here on purpose (unlike ebb's) — that was a deliberate
77
+ choice as far back as Session 9's CLI, not an oversight filled in later.
78
+
79
+ ```
80
+ telltale check <path> --base origin/main
81
+ ```
82
+
83
+ Computes the merge base of the current checkout and `--base` (`git merge-base HEAD <ref>` —
84
+ the commit where they diverged, not `<ref>`'s current tip, which can have moved since a PR
85
+ branched: `src/telltale/git_ref.py`), materializes it into a real temporary `git worktree`
86
+ (so the ordinary gitignore-aware walker sees exactly what a normal checkout would — never a
87
+ bare `git show`/tarball extraction), runs the full pipeline against it, and diffs the two
88
+ matrices (`src/telltale/regression.py`, pure — no I/O, no git). Exit 1 only if something that
89
+ was `present` at the merge base is not `present` now, at either granularity (route-level or a
90
+ specific error mode) — never on an absolute percentage or count. `absent → present` is an
91
+ improvement, not a regression; `ambiguous → absent` (a new extractor started running and found
92
+ nothing) is new information, not a loss; a route added or removed entirely is a spec change,
93
+ not a coverage regression.
94
+
95
+ Handles being run from a subdirectory of a larger repository correctly — `git worktree add`
96
+ always materializes the *whole* repo, never a subtree, so `run_check_at_merge_base` computes
97
+ the offset between the given path and the repo's actual toplevel once (`git_ref.toplevel()`)
98
+ and re-applies it inside the new worktree. An earlier version of this didn't do that and
99
+ happened to pass every test anyway, because every test so far had pointed `--base` straight at
100
+ a repo's own root — caught by testing the monorepo case explicitly, then confirmed against
101
+ this actual repository (`telltale check apps/telltale/tests/fixtures/fixture_service --base
102
+ HEAD`, run for real, not just in pytest) before it shipped.
103
+
104
+ **Slice 4, part 2 (Session 14): `--format`, the composite Action, and the PR comment** — the
105
+ same shape as ebb's own Session 8, roughly half the work, because `packages/keel` and the
106
+ `scripts/upsert_pr_comment.sh` pattern already existed to build on.
107
+
108
+ ```
109
+ telltale check <path> --format table|markdown|annotations
110
+ ```
111
+
112
+ Three renderers now sit behind one flag, all built from the same `Matrix` + `Regression`
113
+ tuple computed once per run (`src/telltale/cli.py`'s `_render`) — never three separate scans:
114
+
115
+ - `table` (default): unchanged from Session 13 — the two Rich tables plus, if `--base` found
116
+ regressions, the regressions table (`src/telltale/render/terminal.py`).
117
+ - `markdown` (`src/telltale/render/markdown.py`, new): the same two tables as plain Markdown
118
+ pipe-tables, plus a "Regressions vs merge base" section when `--base` found any — this is
119
+ the composite Action's PR comment body.
120
+ - `annotations` (`src/telltale/render/annotations.py`, new): GitHub workflow-command syntax
121
+ (`::error file=...,line=...::message`), **one per regression, never one per coverage cell**.
122
+ Unlike ebb's Finding (a model-id string at a specific line), a `CoverageCell` has no file:line
123
+ of its own — the annotation points at the route's OpenAPI definition instead
124
+ (`Regression.route_source_file`/`route_source_line`, taken from the `after` state, i.e.
125
+ HEAD — what a PR reviewer actually has open, not the now-deleted merge-base worktree). A gap
126
+ that already existed at the merge base isn't new information for this PR, so only PRESENT ->
127
+ not-PRESENT gets annotated, same DEC-04 discipline as the exit-code gate itself — there's
128
+ deliberately no "annotate every absent cell" mode.
129
+
130
+ `apps/telltale/action.yml` (composite, not Docker/JS — same reasoning as the repo root's ebb
131
+ action): wraps `uvx foretop-telltale check`, defaults `--base` to the pull request's own base
132
+ SHA, emits `--format annotations` for inline PR annotations (zero extra permissions, same
133
+ mechanism ebb's action already uses) and upserts a `--format markdown` PR comment by HTML
134
+ marker (`apps/telltale/scripts/upsert_pr_comment.sh` — logic duplicated from, not imported
135
+ from, the repo-root copy: docs/adr/0002 requires every app to stay standalone-copyable into
136
+ its own future public mirror, exactly the ebb/`sync-ebb-mirror.sh` precedent, so this app's
137
+ action can't reach outside its own directory for a script it needs). Both copies of the script
138
+ now read their marker from a `COMMENT_MARKER` env var instead of a hardcoded string, so a repo
139
+ running both ebb's and telltale's Actions on the same PR upserts two separate comments instead
140
+ of colliding on one — a small, behavior-preserving generalization made to the shared script,
141
+ not a new one.
142
+
143
+ Lives at `apps/telltale/`, not the repo root (unlike ebb's `action.yml`): a repo can only have
144
+ one `action.yml` per directory, and the root is already ebb's. This also happens to match
145
+ telltale's real state honestly — it has no public mirror yet, so nothing here can assume a
146
+ `uvx foretop-telltale` invocation resolves to a real PyPI release; that's the same not-yet-done
147
+ caveat Session 8 already flagged for ebb's own action before it was ever actually published.
148
+
149
+ Verified for real, the same way Session 13's regression gate was: ran `telltale check
150
+ apps/telltale/tests/fixtures/fixture_service --format markdown` (clean output, 29% coverage),
151
+ then emptied the fixture's `prometheus/rules.yaml` to simulate a real regression and re-ran
152
+ with `--base HEAD --format annotations` (5 correctly-anchored `::error file=openapi.yaml,
153
+ line=N::` lines, exit 1) and again with `--format markdown` (the same 5 rows in a rendered
154
+ "Regressions vs merge base" table), then restored the fixture and confirmed a clean `git
155
+ status` — never just trusted the unit tests alone for the git-worktree-dependent path.
156
+
157
+ **Slice 5 (Session 15): the golden corpus** — `tests/golden/` + `tests/golden_manifest.yaml` +
158
+ `tests/unit/test_golden_corpus.py`, same shape and the same floor as ebb's own (Session 7):
159
+ precision >= 0.95, recall >= 0.85, scored **per extractor** (prometheus/grafana/otel — the
160
+ first path segment under `tests/golden/`, same grouping ebb uses for its own detector dirs),
161
+ never pooled into one aggregate that could hide one broken extractor behind two working ones.
162
+ `make accuracy-telltale` runs just this file, verbosely.
163
+
164
+ Ground truth is one entry per `Signal` a real extractor should produce for a fixture — `path`
165
+ (relative to `tests/golden/`), `evidence_line`, `route_template`, `method`, `status_code`,
166
+ `signal_type`. `null` fields are real ground truth (the extractor has no basis to claim a
167
+ route/method/status), not missing data — see `prometheus/03_no_route_label`, where an alert
168
+ with no route label still produces one Signal, correctly, with `route_template: null`.
169
+
170
+ Every extractor group has at least one pure-decoy fixture — content that shares surface
171
+ vocabulary with a real match without being one — enforced by its own test
172
+ (`test_every_fixture_group_has_a_pure_decoy_case`, identical logic to ebb's own): a 0-FP score
173
+ is trivial if a false positive was never actually possible to produce.
174
+
175
+ **A real bug, found by building this corpus, not just theorized about:** `grafana.py`'s own
176
+ docstring already claimed "a non-Prometheus datasource's query language is out of scope, not
177
+ silently misread as PromQL" — but the code only ever checked for the presence of an `expr`
178
+ field, never the target's actual datasource type. Loki's query field is *also* named `expr`,
179
+ and LogQL's `{label="value"}` stream selector is syntactically close enough to a PromQL
180
+ selector that a Loki panel's own labels (a `route="..."` stream label is entirely realistic)
181
+ would have been misread as real route evidence — confirmed against a real Loki-backed
182
+ dashboard fetched live this session, not assumed. Fixed with `_is_explicitly_non_prometheus`:
183
+ a `datasource.type` naming anything other than `"prometheus"`, on either the panel or the
184
+ target, is skipped. The corpus's own `grafana/04_decoy_loki_panel` fixture is what would have
185
+ caught this, and now guards against it regressing. A narrower gap remains and is left
186
+ documented, not silently patched over: a target with no `datasource` at all, or the legacy
187
+ `"${DS_XXX}"` template-variable shape older exported dashboards use, still reads `expr` as
188
+ PromQL — resolving that would mean cross-referencing the document's own `__inputs`, which is
189
+ more machinery than this parser's "good enough, not semantically perfect" scope calls for.
190
+
191
+ **Session 16: prepping to publish** — the release workflow and public-mirror sync script,
192
+ mirroring `apps/ebb`'s own Session 8 pattern exactly. Not yet actually published: PyPI project
193
+ creation and trusted-publisher configuration were always the user's job, not this session's
194
+ (same NEXT_STEPS.md Track A split ebb's own `docs/adr/0002-ebb-public-mirror.md` already
195
+ documents), and this session stopped short of creating a real `foretop-dev/telltale` repo.
196
+
197
+ - `.github/workflows/telltale-release.yml` (this monorepo) and
198
+ `scripts/mirror-workflows/telltale-{ci,release}.yml` (copied into the mirror as `ci.yml`/
199
+ `release.yml` by `scripts/sync-telltale-mirror.sh`) — same shape as ebb's own three files,
200
+ tag-namespaced `telltale-v*` per `docs/adr/0002`'s already-stated plan for the whole suite.
201
+ - Two real bugs found while building this, not just theorized about:
202
+ - `pyproject.toml`'s `[project.scripts]` only declared `telltale = ...`, not also
203
+ `foretop-telltale = ...` — `uvx <name>` (no `--from`) resolves the entry point matching the
204
+ *distribution* name, so `uvx foretop-telltale check .` (what `apps/telltale/action.yml`
205
+ already promises) would have failed to resolve once published. Caught by actually building
206
+ the wheel and inspecting `entry_points.txt`, not assumed correct because ebb's own
207
+ `pyproject.toml` already has the equivalent fix — confirmed fixed the same way afterward.
208
+ - No `LICENSE` file existed in `apps/telltale/` at all (copied from `apps/ebb/LICENSE`,
209
+ byte-identical standard Apache-2.0 text — a LICENSE file has no per-project content of its
210
+ own); `license-files = ["LICENSE"]` added to `pyproject.toml` so the built wheel actually
211
+ bundles it, verified by inspecting the wheel's own `dist-info/licenses/` after the change.
212
+ - The blocker this section used to describe (`keel.collect.walk`/`keel.git_ref` only resolving
213
+ inside the monorepo's own workspace) is fixed as of a later session — see
214
+ `packages/keel/README.md`'s own "Packaging note": `scripts/sync-telltale-mirror.sh` now
215
+ vendors the exact `keel` submodules telltale imports (`git_ref.py`, `collect/
216
+ line_tracking.py`, `collect/walk.py`) into the mirror at sync time, and the mirror's own
217
+ `pyproject.toml` (`scripts/mirror-pyproject/telltale.toml`) no longer depends on
218
+ `foretop-keel`. Re-verified for real after the fix: `ruff`/`mypy`/`pytest` (156 tests) all
219
+ green standalone, a real wheel built, and the built wheel cold-installed into a fresh venv
220
+ with `telltale --help` actually running.
221
+
222
+ ```
223
+ uv run --package foretop-telltale telltale check <path>
224
+ ```
225
+
226
+ Domain model (`src/telltale/models.py`) is the SUITE_ARCHITECTURE-style pure core:
227
+ `RouteSet` (from the OpenAPI parser) × `SignalSet` (from the signal extractors) →
228
+ `match()` (`src/telltale/match.py`, pure, no I/O) → `Matrix`. Every coverage cell is
229
+ **three-state** — `present` / `absent` / `ambiguous` — never just present/absent. `ambiguous`
230
+ is not a synonym for `absent`: a route gets `ambiguous` for a signal type only when no
231
+ extractor for that type exists in telltale's own code at all — as of Session 11 every
232
+ `SignalType` has a real extractor (Prometheus → `alert`; Grafana → `metric`; OTel Collector →
233
+ `log`/`metric`/`trace`), so `ambiguous` no longer appears in ordinary runs. It stays a real,
234
+ reachable state, not a vestige: the moment a *new* signal dimension is added to `SignalType`
235
+ without an extractor yet, every route goes `ambiguous` for it, honestly, rather than a false
236
+ `absent` — see `SignalSet.attempted_types` in `models.py`. `absent` means an extractor
237
+ genuinely ran across the whole repo and found nothing for that route — specs/telltale.md's
238
+ explicit warning: "never claim coverage that does not exist... ambiguous is its own outcome
239
+ and is never counted as covered."
240
+
241
+ **OpenAPI parser** (`src/telltale/openapi.py`): OpenAPI 3.0/3.1, YAML or JSON. Reads the 8
242
+ Path Item Object method keys (get/put/post/delete/options/head/patch/trace — `spec.openapis.org`
243
+ oas/v3.1.0.html §4.8.7, fetched live this session, not written from memory) and each operation's
244
+ `responses` keys as-is (`"200"`, `"5XX"`, `"default"`) — stored since Session 9, consumed by
245
+ the matcher's error-mode dimension since Session 12. `$ref` resolution is local-document-only
246
+ (`#/components/...` JSON Pointers, walked against the already-parsed document) — external-file
247
+ `$ref`s are out of scope for Slice 1 and produce an `ambiguous`-flavoured skip, never a crash,
248
+ per specs/telltale.md §6's degrade-not-crash rule.
249
+
250
+ **Prometheus rule parser** (`src/telltale/prometheus.py`): parses `groups[].rules[]`
251
+ (`prometheus.io/docs/prometheus/latest/configuration/alerting_rules/`, fetched live), and only
252
+ entries with an `alert:` key — `record:` entries are recording rules, not alerts, and are
253
+ skipped. Extracts route-identifying label matchers from each rule's `expr` with a lightweight
254
+ selector regex (`metric_name{label="value", ...}` — PromQL's own basic matcher syntax,
255
+ `prometheus.io/docs/prometheus/latest/querying/basics/`, fetched live), not a full PromQL
256
+ parser — same "good enough, not semantically perfect" scope decision ebb's detectors made.
257
+ Which label names count as "this is the route" is a **declarative rules file**
258
+ (`src/telltale/rules/label_names.yaml`), not hardcoded `if` branches — specs/telltale.md §5's
259
+ explicit requirement, so this is a config edit for a user extending it, not a code change.
260
+
261
+ **Grafana dashboard parser** (`src/telltale/grafana.py`, Session 11): dashboard JSON
262
+ ("dashboards as code"). Verified against two real, live-fetched public dashboards — a modern
263
+ flat `panels[]` shape and the legacy `rows[].panels[]` shape both use `targets[].expr` for a
264
+ Prometheus-backed panel's query. Only `expr` targets are read (specs/telltale.md DEC-01 commits
265
+ to OTel + Prometheus for v1; a non-Prometheus datasource's query language is out of scope, not
266
+ silently misread). Reuses the exact same PromQL selector/label-matcher logic as the Prometheus
267
+ parser (`src/telltale/promql.py`, extracted in this session so both parsers share one
268
+ implementation) — a panel visualizing a metric produces `metric`-type evidence, the same "this
269
+ is being watched" signal an alert represents for `alert`, without an alerting threshold.
270
+
271
+ **OTel Collector config parser** (`src/telltale/otel.py`, Session 11): `receivers`/
272
+ `processors`/`exporters`/`service.pipelines` — verified against opentelemetry.io's own docs
273
+ live. Deliberately narrow, and says so in its own docstring: most collector configs wire
274
+ generic pipelines with no reference to individual routes at all, so this parser only produces
275
+ evidence where a `filter` processor's OTTL condition actually names a route via `http.route`/
276
+ `http.request.method` (OpenTelemetry's current stable semantic convention attribute names,
277
+ `opentelemetry.io/docs/specs/semconv/http/http-spans/`, fetched live — not the code's own
278
+ `route`/`method` shorthand, since these are dotted attribute names from a different
279
+ ecosystem's convention, so they're a separate section of `label_names.yaml`, not folded into
280
+ the Prometheus lists). A processor only counts if it's genuinely wired into a
281
+ `service.pipelines` entry — an unreferenced `processors:` block is dead config that isn't
282
+ affecting the deployed collector, and reading it would be evidence for something not actually
283
+ happening. Which `traces`/`metrics`/`logs` section a condition sits under determines the
284
+ `SignalType` directly, since a filter processor's own structure already says that
285
+ unambiguously — no need to cross-reference the pipeline for that part.
286
+
287
+ **Evidence, not just data**: every parser needs real line numbers for `evidence_path` /
288
+ `evidence_line` (specs/telltale.md §6: "never assert without evidence"), but plain
289
+ `yaml.safe_load`/`json.loads` discard node positions once a document is built.
290
+ `keel.collect.line_tracking` (originally `src/telltale/yaml_lines.py`, extracted to `keel` in
291
+ Session 15 when charter needed the exact same thing for its own JSON-only config parsers — see
292
+ `packages/keel/README.md`) is a small custom `SafeLoader` that tags every parsed mapping with
293
+ its own starting line — used by every parser here, including the JSON ones (verified directly,
294
+ twice now — first for the OpenAPI parser, then again for a real 12KB Grafana dashboard file —
295
+ that well-formed JSON parses identically through this loader as through `json.loads`, just
296
+ with line tracking available). The one gap: a plain YAML/JSON list of strings (an OTel filter
297
+ condition list) has no per-item position, only the mappings do — `otel.py`'s evidence there
298
+ points at the enclosing `traces:`/`metrics:`/`logs:` block, the finest granularity available,
299
+ not a fabricated per-condition line.
300
+
301
+ **Route-template matching**: an OpenAPI path template (`/users/{id}`) and a signal's claimed
302
+ route (a Prometheus label value, a Grafana panel's query label, an OTel attribute value) are
303
+ compared after normalizing parameter placeholders (`{id}` and `:id` both collapse to the same
304
+ token) — exact match only, no fuzzy scoring, one shared function
305
+ (`src/telltale/route_template.py`) for every extractor. This is deliberately conservative: a
306
+ false `present` is worse than a false `absent` (specs/telltale.md's explicit priority), so an
307
+ unrecognized placeholder convention or a genuinely different path just doesn't match, rather
308
+ than guessing.
309
+
310
+ **Method narrowing**: two operations can share a path template (`GET /todos` and `POST
311
+ /todos`). Route-template matching alone would mark both `present` off a single GET-scoped
312
+ alert — a real false `present`, caught by the fixture-service acceptance test, not just
313
+ reasoned about. When an extractor finds a recognized method label/attribute
314
+ (`label_names.yaml`'s `method_label_names` for Prometheus/Grafana, `otel_method_attribute_names`
315
+ for OTel), a match additionally requires that method to agree with the route's. A signal with
316
+ no method information still matches on route template alone — the best information actually
317
+ available, not a reason to withhold a real match. See `.claude/skills/matching-rules.md` for
318
+ the full reasoning.
319
+
320
+ **The error-mode dimension** (`src/telltale/response_code.py`, Session 12): a route documents
321
+ its response codes as literal codes ("404") or OpenAPI's own wildcard convention ("5XX") —
322
+ `normalize_response_code()` maps both into one comparable space, and `is_error_mode()` restricts
323
+ to 4xx/5xx only (2xx/3xx/1xx are real, normalizable codes but not what DEC-03 tracks). Every
324
+ extractor was extended to also read a status-code label/attribute
325
+ (`label_names.yaml`'s `status_code_label_names` for Prometheus/Grafana,
326
+ `otel_status_code_attribute_names` for OTel) alongside route/method. Matching is **exact string
327
+ equality only, no class-membership reasoning either direction** — a route documenting "5XX" is
328
+ `present` only for a signal also scoped to exactly "5XX", not one scoped to the narrower "503",
329
+ and a route documenting "404" is never satisfied by a signal scoped to the broader "4XX". This
330
+ is a real, deliberate, documented gap (same false-present-is-worse-than-false-absent
331
+ conservatism as method narrowing), not an oversight.
332
+
333
+ One genuine PromQL idiom is recognized as meaningful, not fuzzy-matched: `code=~"5.."` is the
334
+ standard, widely-used convention for "the 5xx class," equivalent to OpenAPI's own "5XX" — only
335
+ this exact two-dot shape is interpreted, nothing broader. OTTL (OTel's condition language) has
336
+ no equivalent single standard idiom for a status-code *range*, so `otel.py` only reads exact
337
+ `==` conditions for status codes — a range condition like
338
+ `attributes["http.response.status_code"] >= 500` is a known, documented gap, not a silent one.
339
+ `http.response.status_code` is int-typed in OTTL (its own semconv example value is unquoted
340
+ `200`) — `otel.py`'s condition regex reads both quoted strings (`http.route`,
341
+ `http.request.method`) and bare integers, verified against a real OTTL condition shape, not
342
+ assumed to be one or the other.
343
+
344
+ **Session 10 (the extraction session):** the gitignore-aware file walker
345
+ (`is_binary`/`iter_files`/the gitignore matching) that `src/telltale/walk.py` copied verbatim
346
+ from `apps/ebb/src/ebb/walk.py` in Session 9 (marked `# DUPLICATED-FROM-EBB`) turned out to be
347
+ byte-identical between the two copies, so it moved to `packages/keel` —
348
+ `keel.collect.walk`. `src/telltale/walk.py` now imports it and keeps only
349
+ `iter_readable_text_files`, telltale's own thin layer on top. `packages/keel/README.md` has
350
+ the full identical-vs-similar reasoning.
351
+
352
+ `src/telltale/brand.py` duplicates ebb's one-line `BRAND` constant rather than importing it,
353
+ per NEXT_STEPS.md's insulation rule — that one stays duplicated; it isn't the kind of thing
354
+ `keel` is for.