fix: prevent fleet snapshot argument overflow

[unhexquadium]

Intent

Fix fm-fleet-view/fm-fleet-snapshot failing with jq 'Argument list too long' at bin/fm-fleet-snapshot.sh:635 once the fleet reaches 17+ task metas, reproducing the current 29-meta failure first. Move all fleet-sized JSON payloads off jq argv and onto stdin or equivalent streamed/file transport while keeping normal-case output byte-identical. Sweep bin/fm-fleet-snapshot.sh and bin/fm-fleet-view.sh for any other jq or xargs call whose argv grows with fleet size and fix those in the same pass. Add a tests/fm-fleet-snapshot-view.test.sh regression using a large synthetic fleet sufficient to trip the old limit. Do not merge the PR. Push the feature branch to the fork over SSH at git@github.com:unhexquadium/firstmate.git, open the PR against upstream kunchenguid/firstmate, and preserve this exact line under a ## Pipeline section in the PR body: Updates from git push no-mistakes. Use a concise top-level summary, a dedicated Risk assessment section with a green/yellow/red status-color emoji, and collapsible details for per-item breakdowns, background, validation notes, and caveats; refer to the captain as she/her. Drive all no-mistakes gates through PR and CI, and when the fork-PR workflow is awaiting upstream maintainer approval after the pipeline finishes, report the PR as delivered with CI awaiting that approval.

What Changed

• Stream task, inventory, secondmate, and final snapshot JSON into jq through stdin instead of passing fleet-sized payloads through --argjson arguments.
• Add a 24-task synthetic fleet regression with large decision payloads that verifies both snapshot generation and fleet view rendering.

Risk Assessment

✅ Low: Captain, the change is a focused transport refactor that moves every fleet-sized JSON value in both snapshot modes off jq argv while preserving the existing JSON structure and rendering path.

Testing

The focused behavior suite passed, the base commit reproduced jq: Argument list too long at line 635 with 29 metas, and the target preserved all 29 tasks through snapshot JSON, secondmate-summary, and rendered-view paths while remaining byte-identical for a normal two-task fleet; the CLI evidence was captured and the worktree remained clean.

Evidence: 29-meta base-versus-target CLI transcript

Source: 29-meta base-versus-target CLI transcript

$ base fm-fleet-snapshot.sh --json with 29 task metas and 40000-byte decision summaries
exit=1
/tmp/fm-large-fleet-e2e.SrGnsW/base-bin/fm-fleet-snapshot.sh: line 635: /usr/bin/jq: Argument list too long
fm-fleet-snapshot: main inventory summary failed

$ target fm-fleet-snapshot.sh --json with the same 29-meta fleet
{
  "schema": "fm-fleet-snapshot.v1",
  "task_count": 29,
  "max_decision_summary_bytes": 40000,
  "main_inventory_valid": true
}

$ target fm-fleet-snapshot.sh --secondmate-home-summary with the same 29-meta fleet
{
  "schema": "fm-secondmate-home-summary.v1",
  "endpoint_count": 29,
  "decision_count": 29,
  "shown_endpoints": 20,
  "omitted": [
    {
      "surface": "decisions_open",
      "count": 9
    },
    {
      "surface": "endpoints",
      "count": 9
    }
  ]
}

$ target fm-fleet-view.sh with the same 29-meta fleet (representative rendered rows)
# Fleet View

Schema: fm-fleet-snapshot.v1
Home: /tmp/fm-large-fleet-e2e.SrGnsW/home

## Under Way
| ID | Current | Kind | Repo/Project | Backend | Endpoint | Artifact | Path | Watch / return channel |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| large-task-01 | unknown / none | ship | alpha | tmux | present | - | - | bin/fm-peek.sh fm-large-task-01 |
| large-task-02 | unknown / none | ship | alpha | tmux | present | - | - | bin/fm-peek.sh fm-large-task-02 |
| large-task-03 | unknown / none | ship | alpha | tmux | present | - | - | bin/fm-peek.sh fm-large-task-03 |
| large-task-29 | unknown / none | ship | alpha | tmux | present | - | - | bin/fm-peek.sh fm-large-task-29 |

$ byte-for-byte normal-case comparison: base vs target with two ordinary task metas
identical=yes
sha256=23d0e11154effa398116f6989c202a4c0b1f0df694ad341276afbb54f61916dd

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

✅ **Review** - passed

✅ No issues found.

✅ **Test** - passed

✅ No issues found.

• tests/fm-fleet-snapshot-view.test.sh
• /home/kalvira/.no-mistakes/evidence/01M0NSSEMQB2TYB1GZFWXJ9RF8/large-fleet-e2e.sh "$PWD" 1231b6ae7fd4c5ff7e94d2f5ca2159536c4c41cb
• rg -n "OUTPUT_MODE|secondmate-home-summary|jq |xargs" bin/fm-fleet-snapshot.sh bin/fm-fleet-view.sh
• git status --short

✅ **Document** - passed

✅ No issues found.

⚠️ **Lint** - 1 warning

• ⚠️ linter found issues (exit code 1)

✅ **Push** - passed

✅ No issues found.

[greptile-apps]

Confidence Score: 5/5

The PR appears safe to merge, with no concrete changed-code failure identified.

The streamed inputs remain valid, positionally stable JSON documents, the refactored field mappings preserve their prior semantics, and the regression exercises the argument-overflow path through both snapshot and view generation.

_{Reviews (1): Last reviewed commit: "Fix fleet snapshot jq argument scaling" | Re-trigger Greptile}