agami-reconcile

Name: agami-reconcile
Author: AgamiAI

19stars0forks

How to install agami-reconcile?

agamiai/agami-core/agami-reconcile

$npx -y skills add AgamiAI/agami-core --skill agami-reconcile

Installs into the current project.

Summary

Reconciles known (label, expected_value) numbers from an existing dashboard against agami's answers. Input can be a SCREENSHOT of a Metabase / Power BI / Tableau / Looker dashboard (Claude's vision extracts the pairs), a CSV, or numbers pasted inline — the user doesn't need to kn

Files · 1

View on GitHub

SKILL.md

1	# agami reconcile
2
3	You are running the reconciliation harness. Goal: take labeled numbers from an existing dashboard (Tableau / Looker / Mode / Metabase / Power BI / spreadsheet) — most often a screenshot, sometimes a CSV or pasted list — and prove agami can reproduce each number. When numbers match, that's evidence the semantic model is right. When they don't, the receipt drill-down explains why — typically a definitional disagreement (gross vs net, refunds in vs out, FX rate at booking vs reporting date) — which is exactly the trust signal that makes a DE relax.
4
5	This skill orchestrates:
6
7	1. Extract the `(label, expected_value)` pairs from the input — a dashboard screenshot (via vision, confirmed with the user), a CSV, or a pasted list. Number parsing is always deterministic (`reconcile.py`).
8	2. Generate a matching NL question for each label.
9	3. Run each question through the same NL→SQL→execute pipeline as agami-query.
10	4. Diff actual vs expected with a tolerance.
11	5. Present a markdown table with per-row status; for mismatches, render the full receipt as a drill-down so the user can find the definitional disagreement.
12
13	Spec for the deterministic helpers: [`scripts/reconcile.py`](../../scripts/reconcile.py) (CSV parser + number normalization + diff with tolerance).
14
15	## Conversation style
16
17	- Tight loops. This skill is a tool, not a tutorial. One question per turn, max two sentences of prose between phases.
18	- Surface mismatches loud. A reconcile run with 9/12 matches and 3 mismatches is a SUCCESSFUL run — the mismatches are the value. Lead with what didn't match.
19	- Don't paste raw SQL in chat. The receipt has it. Same hard rule as agami-query.
20
21	---
22
23	## Phase 0: Preflight
24
25	Same checks as agami-query / agami-connect:
26
27	1. Plan-mode check per [`shared/plan-mode-check.md`](../../shared/plan-mode-check.md). This skill needs Bash + Read + Write — refuse if locked in plan mode. DO NOT write a plan file. DO NOT call `ExitPlanMode`. Refusal text: "I can't reconcile in plan mode — each row runs a live query and writes a receipt. Switch to Auto* or Edit Automatically mode (Shift+Tab to cycle) and re-invoke me with the CSV path."*
28	2. Credentials present — read `<artifacts_dir>/local/credentials` for the active profile. If missing, invoke `/agami-connect` to set up first; this skill needs a working DB connection.
29	3. Model present — `<artifacts_dir>/<profile>/org.yaml` must exist. If not, invoke `/agami-connect`. This skill needs an introspected model to generate questions against.
30	4. Input — accept any of three shapes; the user needn't know which. Detect what they gave:
31	- A screenshot / image of a dashboard (Metabase, Power BI, Tableau, Looker, a spreadsheet) — the common case. Go to Phase 1's vision branch.
32	- A CSV — a path in `$ARGUMENTS`, or pasted inline (write inline CSV to `/tmp/agami-reconcile-<ts>.csv`). Go to Phase 1's CSV branch.
33	- Numbers pasted inline as a list/table — treat as inline CSV.
34	If they gave nothing (or just asked "can you check my dashboard?"), ask once, welcoming all three: "Show me the numbers you want to check against — easiest is a screenshot of your dashboard* (Metabase, Power BI, Tableau, a spreadsheet — whatever you have), but a CSV or a pasted list of `label: value` works too."* Don't make them figure out an export format.
35	5. If they gave a file path, validate it exists. If not, surface the error and stop.
36
37	---
38
39	## Phase 1: Extract the (label, value) pairs
40
41	Whatever the input shape, the goal is the same normalized rows JSON. *

People who liked this also viewed

6 picks

Type

agami-reconcile

19stars0forks

View on GitHub

How to install agami-reconcile?

agamiai/agami-core/agami-reconcile

$npx -y skills add AgamiAI/agami-core --skill agami-reconcile

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# agami reconcile
2
3	You are running the reconciliation harness. Goal: take labeled numbers from an existing dashboard (Tableau / Looker / Mode / Metabase / Power BI / spreadsheet) — most often a screenshot, sometimes a CSV or pasted list — and prove agami can reproduce each number. When numbers match, that's evidence the semantic model is right. When they don't, the receipt drill-down explains why — typically a definitional disagreement (gross vs net, refunds in vs out, FX rate at booking vs reporting date) — which is exactly the trust signal that makes a DE relax.
4
5	This skill orchestrates:
6
7	1. Extract the `(label, expected_value)` pairs from the input — a dashboard screenshot (via vision, confirmed with the user), a CSV, or a pasted list. Number parsing is always deterministic (`reconcile.py`).
8	2. Generate a matching NL question for each label.
9	3. Run each question through the same NL→SQL→execute pipeline as agami-query.
10	4. Diff actual vs expected with a tolerance.
11	5. Present a markdown table with per-row status; for mismatches, render the full receipt as a drill-down so the user can find the definitional disagreement.
12
13	Spec for the deterministic helpers: [`scripts/reconcile.py`](../../scripts/reconcile.py) (CSV parser + number normalization + diff with tolerance).
14
15	## Conversation style
16
17	- Tight loops. This skill is a tool, not a tutorial. One question per turn, max two sentences of prose between phases.
18	- Surface mismatches loud. A reconcile run with 9/12 matches and 3 mismatches is a SUCCESSFUL run — the mismatches are the value. Lead with what didn't match.
19	- Don't paste raw SQL in chat. The receipt has it. Same hard rule as agami-query.
20
21	---
22
23	## Phase 0: Preflight
24
25	Same checks as agami-query / agami-connect:
26
27	1. Plan-mode check per [`shared/plan-mode-check.md`](../../shared/plan-mode-check.md). This skill needs Bash + Read + Write — refuse if locked in plan mode. DO NOT write a plan file. DO NOT call `ExitPlanMode`. Refusal text: "I can't reconcile in plan mode — each row runs a live query and writes a receipt. Switch to Auto* or Edit Automatically mode (Shift+Tab to cycle) and re-invoke me with the CSV path."*
28	2. Credentials present — read `<artifacts_dir>/local/credentials` for the active profile. If missing, invoke `/agami-connect` to set up first; this skill needs a working DB connection.
29	3. Model present — `<artifacts_dir>/<profile>/org.yaml` must exist. If not, invoke `/agami-connect`. This skill needs an introspected model to generate questions against.
30	4. Input — accept any of three shapes; the user needn't know which. Detect what they gave:
31	- A screenshot / image of a dashboard (Metabase, Power BI, Tableau, Looker, a spreadsheet) — the common case. Go to Phase 1's vision branch.
32	- A CSV — a path in `$ARGUMENTS`, or pasted inline (write inline CSV to `/tmp/agami-reconcile-<ts>.csv`). Go to Phase 1's CSV branch.
33	- Numbers pasted inline as a list/table — treat as inline CSV.
34	If they gave nothing (or just asked "can you check my dashboard?"), ask once, welcoming all three: "Show me the numbers you want to check against — easiest is a screenshot of your dashboard* (Metabase, Power BI, Tableau, a spreadsheet — whatever you have), but a CSV or a pasted list of `label: value` works too."* Don't make them figure out an export format.
35	5. If they gave a file path, validate it exists. If not, surface the error and stop.
36
37	---
38
39	## Phase 1: Extract the (label, value) pairs
40
41	Whatever the input shape, the goal is the same normalized rows JSON. *

People who liked this also viewed

6 picks

Type