feat(jtk): add issue history command

[rianjs]

Summary

• Add Jira changelog API support and jtk issues history <issue-key>.
• Render compact changed-field history rows with --id, --extended, --fields, --max, and --next-page-token.
• Update README/output spec/changelog and bump tools/jtk/version.txt to 1.3.

Verification

• go test ./api ./internal/present ./internal/cmd/issues
• make check

Live Validation

• Built the local binary with make build.
• Validated configured credentials with bin/jtk --non-interactive me.
• Validated read-only issue discovery with bin/jtk --non-interactive issues list --project MON --max 20 --id.
• Validated bin/jtk --non-interactive issues history MON-4648 --max 5.
• Validated bin/jtk --non-interactive issues history MON-4648 --id --max 5.
• Validated bin/jtk --non-interactive issues history MON-4648 --extended --max 5.
• Validated bin/jtk --non-interactive issues history MON-4648 --fields CREATED,FIELD,TO --max 5.
• Validated pagination with bin/jtk --non-interactive issues history MON-4648 --max 1 --next-page-token 1; the first page returned next: 1, and the second returned next: 2.

Closes #420

[rianjs]

Findings

Major: live read-only validation is still unmet.
PR body / Live Validation documents a credential/keyring failure, but the original batch requirement and converged plan required a locally built bin/jtk to exercise jtk issues history against configured Jira credentials, including forced --max 1 pagination. Since this is a new Jira endpoint wrapper, the most important integration behavior remains unverified. Before merge, either run the live read-only matrix after fixing local auth, or record an explicit maintainer waiver for this ticket.

Minor: command-level coverage does not exercise --extended or --fulltext.
history_test.go covers default output, --id, projection, bad token, and empty pages, while presenter tests cover extended/fulltext directly. Add one command-path test proving opts.Extended produces extended columns and disables truncation, or opts.FullText disables truncation, so the global read-flag contract is locked at the command boundary.

Nit: history README flag table omits --extended.
README.md includes an --extended example but the command-specific flag table lists --fulltext and --id only. Add --extended there for consistency.

Nit: OUTPUT_SPEC text omits TYPE from the extended-only list.
OUTPUT_SPEC.md lists several extended-only columns but skips TYPE, even though it appears in the extended header. Small doc correction.

[rianjs]

Live read-only validation with the local build is now complete.

Target: MON-4648 using configured credentials from this machine.

Validated commands:

• bin/jtk --non-interactive me
• bin/jtk --non-interactive issues list --project MON --max 20 --id
• bin/jtk --non-interactive issues history MON-4648 --max 5
• bin/jtk --non-interactive issues history MON-4648 --id --max 5
• bin/jtk --non-interactive issues history MON-4648 --extended --max 5
• bin/jtk --non-interactive issues history MON-4648 --fields CREATED,FIELD,TO --max 5
• bin/jtk --non-interactive issues history MON-4648 --max 1 --next-page-token 1

Results: all commands returned successfully. Pagination was verified from next: 1 to the second history row and returned next: 2.

[rianjs]

Findings

Minor: PR body still records the old failed live-validation state.
The current PR body’s Live Validation section still says configured-credential validation was blocked by keyring/auth state, while the updated validation state says local built bin/jtk passed against MON-4648 for default, --id, --extended, --fields CREATED,FIELD,TO, and --next-page-token. Please update the PR body or add a prominent validation summary that points to the passing comment, so the merge record does not preserve the earlier failed-validation state.

No code-level architectural findings after the follow-up. The prior non-live findings are addressed: command-level extended/fulltext behavior is covered, --extended is documented, TYPE is included in the extended-only output spec list, and the live-validation requirement is satisfied based on the provided validation state.

[rianjs]

Findings

No findings.

The remaining PR-body validation issue is resolved. The current metadata now records successful configured-credential validation with the local build, including default output, --id, --extended, projected fields, and token pagination. The diff remains aligned with the ticket intent: compact changelog rows, standard read flags, bounded output, projection identity behavior, docs, tests, and jtk minor version bump.

[rianjs]

Findings

• Major: The new history tests do not exercise the full Cobra path for the headline flags. history_test.go covers --fields via cmd.Execute(), but --id, --extended, --fulltext, and a successful non-empty --next-page-token are only covered through direct runHistory(...) calls or presenter tests. That means a broken flag binding or RunE handoff could still pass CI even though the public command surface is wrong. I’d add one end-to-end cmd.Execute() case that uses a real page token and at least one of the global flags, then assert both the request startAt and the rendered mode.
• Minor: All fixtures use a single changelog group, so the suite never proves that multiple history groups stay in Jira order or that group IDs are preserved/repeated correctly across a page. A regression that reorders or drops groups would still pass today. Add a two-group fixture in history_test.go or issue_history_test.go.

[monit-reviewer]

Automated PR Review

Reviewed commit: f1ddf61

Summary

Reviewer	Findings
documentation:docs-reviewer	2
harness-engineering:harness-self-documenting-code-reviewer	3

documentation:docs-reviewer (2 findings)

💡 Suggestion - tools/jtk/CHANGELOG.md:20

New entry links to the GitHub issue (#420 via issues/420) while all other CHANGELOG entries consistently link to pull requests (pull/327, pull/189). GitHub auto-redirects so the link works, but the inconsistency may confuse readers tracing history.

💡 Suggestion - tools/jtk/internal/cmd/OUTPUT_SPEC.md:316

--id output mode is described in prose but has no rendered example block, unlike the default and --extended modes which both have examples. Readers cannot verify what --id output actually looks like from the spec alone.

harness-engineering:harness-self-documenting-code-reviewer (3 findings)

💡 Suggestion - tools/jtk/internal/cmd/issues/history.go:106

issueHistoryPagination is a noun phrase that does not surface the return contract (bool, string) → (hasMore, nextToken). A verb-led name such as computeHistoryPageCursor or nextHistoryPageToken would let a reader understand what the function produces without scanning the body or call site.

⚠️ Should Fix - tools/jtk/internal/cmd/issues/history.go:111

The guard if advance == 0 && page.MaxResults > 0 { advance = page.MaxResults } handles a non-obvious Jira API edge case — an empty page mid-set — with no indication of why removing it would reintroduce a pagination loop. A short why-comment (e.g., // guard: avoid re-emitting the current offset when the API returns an empty page mid-set) is warranted here per the project standard of commenting non-obvious invariants.

💡 Suggestion - tools/jtk/internal/present/issue_history.go:154

s = strings.Join(strings.Fields(s), " ") is non-obvious without context: Jira changelog values can embed newlines and tab characters that would corrupt tab-separated table output. A short inline comment such as // collapse embedded newlines/tabs from Jira values explains the constraint to a future reader.

5 PR discussion threads considered.

---

Completed in 4m 05s | $2.40 | sonnet | daemon 0.2.127+dev.1.2b1249f | Glorfindel

Field	Value
Model	sonnet
Reviewers	hybrid-synthesis, documentation:docs-reviewer, harness-engineering:harness-architecture-reviewer, harness-engineering:harness-enforcement-reviewer, harness-engineering:harness-knowledge-reviewer, harness-engineering:harness-self-documenting-code-reviewer, security:security-code-auditor
Engine	claude · sonnet
Reviewed by	pr-review-daemon · monit-pr-reviewer
Duration	4m 05s wall · 4m 02s compute (Reviewers: 2m 58s · Synthesis: 1m 04s)
Cost	$2.40 (estimated)
Tokens	403.5k in / 23.7k out
Turns	14

Per-workstream usage

Workstream	Model	In	Out	Cache read	Cache create	Cost
hybrid-synthesis	sonnet	39.8k	2.9k	13.1k	26.6k (1h)	$0.21
documentation:docs-reviewer	sonnet	39.0k	2.7k	10.6k	28.4k (1h)	$0.21
harness-engineering:harness-architecture-reviewer	sonnet	63.6k	954	0	63.6k (1h)	$0.40
harness-engineering:harness-enforcement-reviewer	sonnet	65.2k	2.1k	10.6k	54.6k (1h)	$0.36
harness-engineering:harness-knowledge-reviewer	sonnet	64.4k	1.5k	10.6k	53.8k (1h)	$0.35
harness-engineering:harness-self-documenting-code-reviewer	sonnet	71.8k	12.3k	10.6k	61.3k (1h)	$0.56
security:security-code-auditor	sonnet	59.8k	1.2k	10.6k	49.2k (1h)	$0.32

_{Re-reviews only run when @monit-reviewer is re-requested as a reviewer — push as many commits as you need, then re-request when ready. PRs targeting branches other than main, master are skipped, even when @monit-reviewer is re-requested.}

Superseded by updated review

[monit-reviewer]

Automated PR Review

Reviewed commit: 09e3cc3 | Previous: f1ddf61 (incremental)

Summary

No issues found.

10 PR discussion threads considered.

---

Completed in 1m 09s | $6.05 | sonnet | daemon 0.2.127+dev.1.2b1249f | Glorfindel

Field	Value
Model	sonnet
Mode	Re-review · Cycle 2 · Session resumed
Reviewers	hybrid-synthesis, documentation:docs-reviewer, harness-engineering:harness-architecture-reviewer, harness-engineering:harness-enforcement-reviewer, harness-engineering:harness-knowledge-reviewer, harness-engineering:harness-self-documenting-code-reviewer, security:security-code-auditor
Engine	claude · sonnet
Reviewed by	pr-review-daemon · monit-pr-reviewer
Duration	1m 09s wall · 44s compute (Reviewers: 28s · Synthesis: 16s)
Cost	$6.05 (estimated)
Tokens	1063.2k in / 26.1k out
Turns	28

Per-workstream usage

Workstream	Model	In	Out	Cache read	Cache create	Cost
hybrid-synthesis	sonnet	90.5k	3.4k	26.3k	64.2k (1h)	$0.44
documentation:docs-reviewer	sonnet	89.7k	2.8k	23.7k	66.0k (1h)	$0.45
harness-engineering:harness-architecture-reviewer	sonnet	164.0k	1.1k	13.1k	150.9k (1h)	$0.93
harness-engineering:harness-enforcement-reviewer	sonnet	168.7k	2.3k	23.7k	145.0k (1h)	$0.91
harness-engineering:harness-knowledge-reviewer	sonnet	166.4k	1.7k	23.7k	142.7k (1h)	$0.89
harness-engineering:harness-self-documenting-code-reviewer	sonnet	188.4k	12.5k	23.7k	164.7k (1h)	$1.18
security:security-code-auditor	sonnet	152.4k	1.4k	23.7k	128.6k (1h)	$0.80
discussion-summarizer	—	43.0k	719	0	43.0k (1h)	$0.45

_{Re-reviews only run when @monit-reviewer is re-requested as a reviewer — push as many commits as you need, then re-request when ready. PRs targeting branches other than main, master are skipped, even when @monit-reviewer is re-requested.}

[rianjs]

Findings

No findings.

The daemon edits are consistent with the approved architecture: docs are clearer, --id output is now explicitly specified, the pagination helper name is more precise, and the comments explain real guards without changing behavior. The added tests strengthen the command-boundary and ordering coverage without expanding scope.