Skip to content
Agentmart
Submit a tool

reconify-engine-debug

Diagnose unexpected Reconify results.

Skill by reconifyhq

IndexedMIT★ 8
Write a review

Chalk Labs

Need help deploying reconify-engine-debug?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/reconifyhq-reconify-engine-debug

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Diagnose unexpected Reconify results. Use when counters, exception events, duplicates, or grouped matches disagree with the expected business scenario.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/reconify-engine-debug

What's inside

Skill name
reconify-engine-debug
Description
Diagnose unexpected Reconify results. Use when counters, exception events, duplicates, or grouped matches disagree with the expected business scenario.
License
MIT
Instructions
72 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Reconify Engine Debug

Trace the first unsupported assumption; preserve the original config and result as evidence.

## 1. Gather first-line evidence

```bash
reconify verify --config reconify.yaml --pair PAIR
```

Read the failing checks before reading any events. `result_reproducible` failing means the retained
`result.json` is stale or the run is not deterministic: regenerate it before trusting its counters.
`explanation_consistent` failing means `explanation.json` does not describe that result: regenerate
it. A `source_check` failure points at the mapping or `file_pattern` that no longer fits the file.
When every check passes, the artifacts are sound and the cause is in the mapping or policy.

## 2. Read the result contract and explanation

```bash
reconify schema result
reconify explain RESULT > explanation.json
```

For NDJSON, inspect individual events as well as the final summary. This checkpoint is complete when
the unexpected counter is tied to concrete event rows and source names.

## 3. Trace symptom to cause

| Symptom | First evidence to test |
|---|---|
| Everything unmatched | Both `ref_col` mappings contain the same identifier domain; both date layouts parse |
| Similar `amount_diff` values | A documented fee or FX policy explains the magnitude |
| `financial_effect_diff` events | The mapped field, expectation form, percentage base/rate, tolerance, and sign operation explain the difference |
| `settlement_diff` events | Gross, net, and configured effect components do not satisfy the declared settlement identity |
| `financial_unchecked` events | The field is mapped but no expectation is configured; treat it as informational unless the business requires a rule |
| Amounts differ by a factor of 100 | Both source multipliers express the same minor unit |
| One- or two-day `timing_diff` events | A documented settlement delay explains the offset |
| Duplicate events | The mapped reference is unique, or the config intentionally groups repeats |
| Ambiguous groups | The chosen group key and pass express the real cardinality |
| Empty names | Each parser maps the intended name column |

This checkpoint is complete when one mapping, parser rule, pass, or business policy explains the
observed rows.

## 4. Prove the correction

Reduce the inputs to the smallest fixture that retains the symptom. Correct the mapping or parser
first; change a tolerance only after identifying the exact rows it would newly absorb and confirming
they are the same transaction.

Then rerun:

```bash
reconify config validate --config reconify.yaml
reconify config check-source --config reconify.yaml --source SOURCE --file INPUT
reconify reconcile --config reconify.yaml --pair PAIR --format json \
  --result-mode all --deterministic --out corrected-result.json
reconify explain corrected-result.json > corrected-explanation.json
```

Debugging is complete when the reduced fixture proves the correction, the full run preserves all
expected counters, and the report names the causal config change. A lower match rate is acceptable;
silencing an unverified financial difference is not.

Financial differences are independent of ordinary matching. Do not widen `amount_tolerance_minor` to
hide a fee or settlement discrepancy; correct the financial expectation or source mapping instead.

When the correction ends the task, promote the corrected artifacts to `result.json` and
`explanation.json`, require `reconify verify` to pass, and report them using the completion
checklist in `../reconify-engine-reconcile/SKILL.md`.

From reconifyhq/reconify, MIT.

Signals

Stars
8
Forks
3
Open issues
13
Last commit (this folder)
Oct 7, 2026 (22 days ago)
Last release
Aug 29, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from reconifyhq

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get reconify-engine-debug running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying reconify-engine-debug?

CallGet help
Cite this page

reconify-engine-debug skill. Agent Mart. https://agentmart.dev/skills/reconifyhq-reconify-engine-debug (last updated 2026-10-24)