dr-status

You are displaying a research progress dashboard. Read the project files and present a clear status overview.

Steps

1. Task Counts

Read todo.md and count:

• [x] — completed tasks
• [!] — failed tasks
• [ ] — remaining tasks
• Total tasks

Break down by phase (P1.xx, P2.xx, etc.).

2. Phase Status

Read .research/ROADMAP.md and show each phase with its status (NOT_STARTED, IN_PROGRESS, COMPLETE).

3. Data Completeness

For each file in data/:

• Count total entries
• For each required schema field, calculate the percentage that are NOT "NOT_FOUND", null, or empty
• Flag fields with >30% NOT_FOUND rate
• Note any CONFLICTING or UNVERIFIED values

4. Recent Activity

Show the last 10 entries from .research/CHANGELOG.md.

5. Trust Score

Read all sidecar files (data/*.adversary.json). For each sidecar file:

• Parse the JSON array of adversary verdicts
• Error handling: If a sidecar file contains malformed JSON, skip it, note "[WARNING] Malformed sidecar: {filename}" in the dashboard output, and continue processing other files
• Count verdicts by type: confirmed, refuted, weakened, unverifiable, and count claims still pending (claims with provenance but no matching sidecar verdict)

Calculate:

• adversary_survival_rate: (confirmed + weakened) / (confirmed + weakened + refuted). Exclude pending and unverifiable from denominator.
• mean_survival_score: average survival_score across all adversary-reviewed claims (those with verdict != pending)
• pending_count: claims with provenance envelopes but no adversary verdict yet
• total_reviewed: confirmed + weakened + refuted + unverifiable

Find the top 5 most-attacked entities by count of refuted + weakened verdicts. For each, find the claim with the lowest survival_score.

Display logic:

• If no sidecar files exist AND provenance exists in data files → show Trust: -- (adversary not yet run)
• If no sidecar files exist AND no provenance exists → show nothing (trust section omitted)
• If --no-adversary was used (detected by: provenance exists in data files but zero sidecar files AND the CHANGELOG contains an entry with Adversary: disabled) → show Adversary: disabled
• If sidecar files exist → show full trust dashboard

6. Dashboard Output

Print a formatted dashboard:

## Research Progress

Phase | Status      | Done | Failed | Remaining
------|-------------|------|--------|----------
P1    | COMPLETE    |   18 |      2 |         0
P2    | IN_PROGRESS |    5 |      1 |        14
P3    | NOT_STARTED |    0 |      0 |        20
P4    | NOT_STARTED |    0 |      0 |        25

Total: 23 done, 3 failed, 59 remaining (85 total)
Overall: 27% complete

## Data Completeness

File                    | Entries | Completeness
------------------------|---------|-------------
data/companies-p1.json  |      18 | 82%
data/companies-p2.json  |       5 | 75%

Fields below 70%: funding_amount (45%), employee_count (62%)

## Trust Score

Trust: 87% survived | Score: 0.74
Adversary: 45/52 claims reviewed (7 pending)

## Most Attacked

Entity           | Refuted | Weakened | Lowest claim
-----------------|---------|----------|-----------------------------
Regie.ai         |       2 |        1 | "funding: $30M" (0.12)
Orum             |       0 |        2 | "employee_count: 60" (0.42)

## Recent Activity

[last 10 changelog entries]

7. Suggest Next Action

Based on the state:

• Unchecked tasks remain → "Run /dr-run to continue execution."
• Phase just completed → "Run /dr-review to validate data quality, then /dr-improve to optimize the researcher."
• All phases complete → "Run /dr-report to synthesize findings."
• High failure rate (>30%) → "Run /dr-improve to analyze failures and improve the researcher."

Error Handling

• If todo.md is missing, tell the user to run /dr-new first.
• If no data files exist yet, show task counts and phases only.