@rainnworks/indy

Give AI agents full access to French accounting — categorize transactions, file VAT, upload receipts, read financial statements — all through Indy.fr.

What is Indy?

Indy.fr is a French accounting SaaS platform used by tens of thousands of freelancers, consultants, and small companies (IS/BIC regime). It handles:

• Daily bookkeeping — bank transaction sync, categorization against the Plan Comptable Général (PCG), receipt attachment
• Tax declarations — monthly/quarterly VAT (CA3/CA12), corporate tax advances (2571), DAS2 (fees), annual closing (AG/PV)
• Financial reporting — balance sheet (bilan), income statement (compte de résultat), trial balance (balance générale)
• Client invoicing — create, track, and reconcile client invoices

All of this happens through a web app at app.indy.fr. There is no official public API.

Why does this exist?

Indy's web interface is designed for humans clicking through forms. But accounting is repetitive — categorizing 50 bank transactions, checking VAT rates, uploading receipt photos — and AI agents are good at repetitive tasks when given the right tools.

This project reverse-engineers Indy's internal API (discovered via MITM traffic analysis) and wraps every useful endpoint as a standalone executable tool that any AI agent can call. Combined with structured skill files that teach agents when and how to use each tool, you get an AI accountant that can do in seconds what takes a human 30 minutes.

What is this?

@rainnworks/indy is an OpenClaw plugin that provides:

• 65 standalone tool binaries — each a self-contained ~27KB Bun script. No framework needed at runtime. Pass JSON params, get JSON back.
• Structured skill files — markdown documents organized as parent routers + detail sub-files that teach agents the workflow logic (decision trees, cross-workflow jumps, edge cases).
• OS-level security model — when creating an agent, you copy only the binaries and skill files it needs into its workspace. If a binary isn't in the folder, the agent can't call it. No allowlists to misconfigure.
• CLI — human-friendly commands for quick lookups (openclaw indy transactions, openclaw indy dashboard).

How it fits together

┌─────────────────────────────────────────────────┐
│  Source: openclaw-indy/                         │
│    bin/          ← 65 standalone tool binaries  │
│    skills/indy/  ← 18 skill files (7 parents   │
│                    + 9 sub-files + SKILL.md     │
│                    + context.md)                │
│    src/          ← TypeScript source            │
└──────────────┬──────────────────────────────────┘
               │  cp only what you need
               ▼
┌─────────────────────────────────────────────────┐
│  Agent workspace: ~/agents/my-agent/            │
│    tools/   ← subset of binaries for this job   │
│    skills/  ← subset of skill files for this job│
└─────────────────────────────────────────────────┘

Two Ways to Use It

Option A: Standalone Binaries (recommended for custom agents)

The repo ships pre-built binaries in bin/ and compiled JS in dist/ — no build step needed. Clone and go.

git clone https://github.com/rainnworks/openclaw-indy.git
cd openclaw-indy

Each binary is a self-contained Bun script (~27KB). Pass INDY_REFRESH_TOKEN as an env var and optional JSON params as the first argument:

# Check how many transactions need categorizing
INDY_REFRESH_TOKEN="..." ./bin/indy-categorize-count.js

# Get the fiscal dashboard
INDY_REFRESH_TOKEN="..." ./bin/indy-get-dashboard.js '{"year":"2025"}'

# Search transactions
INDY_REFRESH_TOKEN="..." ./bin/indy-list-transactions.js '{"search":"stripe","page":1}'

To create an agent, copy only the binaries + skill files it needs into its workspace (see Creating Agents below).

Option B: OpenClaw Plugin

Install as a managed OpenClaw plugin. All 65 tools register automatically — read tools are available by default, write tools require opt-in via the agent allowlist.

openclaw plugins install @rainnworks/indy

Add your token to the OpenClaw config:

plugins:
  entries:
    indy:
      enabled: true
      config:
        refreshToken: "<your-token>"

Enable write tools per-agent:

{
  agents: {
    list: [{
      id: "main",
      tools: {
        allow: [
          "indy_set_accounting_account",   // specific tool
          "indy",                           // all tools from this plugin
          "group:plugins",                  // all plugin tools
        ]
      }
    }]
  }
}

The plugin ships skills at skills/indy/ which load automatically when the plugin is enabled.

Getting your refresh token

There is no official Indy API — this uses Indy's internal API via your session token.

1. Open app.indy.fr, log in, open browser console (F12) and paste:

(async()=>{const r=indexedDB.open('firebaseLocalStorageDb');r.onsuccess=()=>{const tx=r.result.transaction('firebaseLocalStorage','readonly');tx.objectStore('firebaseLocalStorage').getAll().onsuccess=async(e)=>{const t=e.target.result[0]?.value?.stsTokenManager?.refreshToken;if(!t){console.error('No refresh token found. Are you logged in?');return}try{await navigator.clipboard.writeText(t);console.log('✅ Refresh token copied to clipboard!')}catch{console.log('⚠️  Could not auto-copy. Here is your refresh token:\n\n'+t)}}}})()

2. Verify it works:

INDY_REFRESH_TOKEN="<your-token>" ./bin/indy-get-user.js

Rebuilding (optional)

The bin/ and dist/ directories are committed to the repo — you only need to rebuild if you modify the source.

bun install
bun run build.ts          # regenerate bin/ (65 standalone binaries)
npx tsc                   # regenerate dist/ (compiled JS for OpenClaw plugin)

Tool Inventory (65 tools)

Auth & Profile

| Tool | Description | |------|-------------| | indy-login | Setup instructions or verify a token | | indy-get-user | User profile: name, email, company, SIRET, regime, plan | | indy-get-payment | Subscription and billing info |

Transaction Categorization (core daily workflow)

| Tool | Description | |------|-------------| | indy-list-pending-transactions | Uncategorized transactions queue | | indy-categorize-count | Count of transactions needing categorization | | indy-get-similar-transactions | Find previously categorized matches for a transaction | | indy-get-similar-transactions-to-split | Find split patterns for multi-category transactions | | indy-get-category-groups | All accounting categories (PCG codes) | | indy-set-accounting-account | ⚠️ Set the accounting category on a transaction subdivision | | indy-set-transaction-vat | ⚠️ Set VAT rate (domestic, non-EU/intracom, or to-be-defined) | | indy-annotate-transaction | ⚠️ Add/update a note on a transaction | | indy-get-transaction | Get a single transaction by ID | | indy-list-transactions | Search/filter the full ledger | | indy-get-transaction-filters | Valid filter values (accounts, categories, date ranges) | | indy-get-matching-invoices | Find client invoices matching transaction IDs |

Receipts

| Tool | Description | |------|-------------| | indy-get-receipts | Receipts attached to a transaction | | indy-get-receipt | Single receipt by ID (with OCR data) | | indy-upload-receipt | ⚠️ Attach a receipt to a transaction (file path or base64) | | indy-upload-standalone-receipt | ⚠️ Upload a receipt without attaching to a transaction | | indy-list-unpaired-receipts | Receipts not linked to any transaction | | indy-get-receipt-matching-transactions | Find bank transactions matching a receipt | | indy-get-matching-receipts | Find receipts matching a list of transaction IDs | | indy-update-receipt-data | ⚠️ Correct OCR-extracted receipt metadata |

Manual Transactions

| Tool | Description | |------|-------------| | indy-get-manual-transaction-types | Available types: professionalExpense, cashIncome, etc. | | indy-create-manual-transaction | ⚠️ Create an expense note, cash transaction, or personal income | | indy-create-transaction-from-receipt | ⚠️ Create a transaction using a receipt's OCR data | | indy-delete-manual-transaction | ⚠️ Delete a manually created transaction |

VAT & Closing

| Tool | Description | |------|-------------| | indy-get-vat-config | VAT regime: hasVat, frequency, liability type | | indy-list-activities | All fiscal obligations with deadlines and status | | indy-refresh-closing | Force-refresh closing state for a declaration period | | indy-get-closing-state | Read-only closing state (faster, no refresh) | | indy-get-categorize-all-status | Uncategorized transactions blocking a closing | | indy-get-vat-rate-verification | Transactions with undefined VAT in a closing period | | indy-set-closing-step-status | ⚠️ Mark a closing step as done | | indy-get-vat-company-identity | Company identity for a VAT declaration | | indy-set-vat-company-identity | ⚠️ Update company identity (SIRET, IBAN, address) | | indy-get-vat-declaration-summary | CA3/CA12 declaration lines with amounts | | indy-validate-vat-declaration-summary | ⚠️ Confirm the declaration summary step | | indy-get-vat-credit-refund | VAT credit refund eligibility and amounts | | indy-validate-vat-credit-refund | ⚠️ Confirm the credit refund step | | indy-get-vat-validation | Pre-submission validation: declaration type, teletransmission flag | | indy-get-pending-teletransmission | Check for pending teletransmissions | | indy-submit-vat-teletransmission | ⚠️ IRREVERSIBLE — submit declaration to tax authorities | | indy-get-last-teletransmission-state | Last transmission status |

Financial Statements

| Tool | Description | |------|-------------| | indy-get-balance-sheet | Bilan (actif/passif) for a fiscal year | | indy-get-income-statement | Compte de résultat (P&L) for a fiscal year | | indy-get-general-balance | Trial balance: per-account debit/credit/balance | | indy-get-period-choices | Available fiscal periods for documents |

Company & Associates

| Tool | Description | |------|-------------| | indy-get-company-settings | Full settings: regime, legal form, SIREN, APE, address, VAT config | | indy-get-associates | Shareholders: name, birth, shares (‰), entry/exit dates |

Activity Details (Annual Obligations)

| Tool | Description | |------|-------------| | indy-get-cfe-declaration | CFE declaration status and steps | | indy-get-das2 | DAS2 (fees to third parties) status | | indy-get-is-advance-payment | IS advance payment (2571) status and teletransmission | | indy-get-ag-pv | Assemblée Générale / Procès-Verbal status | | indy-get-vehicle-taxes | Vehicle taxes declaration status |

Client Invoicing

| Tool | Description | |------|-------------| | indy-list-client-invoices | Client invoices by status (PENDING, PAID, LATE) | | indy-list-draft-invoices | Draft invoices | | indy-get-invoice-kpis | Invoicing KPIs: invoiced, paid, late, waiting (TTC/HT) | | indy-list-products | Product catalog for invoicing |

Other

| Tool | Description | |------|-------------| | indy-get-dashboard | Monthly income/expense/VAT breakdown by year | | indy-get-accounting-summary | Quick totals: count, income, expenses | | indy-list-bank-accounts | Connected bank accounts with balances and sync status | | indy-list-documents | Stored accounting documents | | indy-list-invoices | Indy subscription billing invoices | | indy-fetch | ⚠️ Raw authenticated API call (escape hatch) |

Tools marked ⚠️ are writes — always confirm with the user before calling them.

Skills Architecture

The plugin ships structured skill files in skills/indy/ that teach an agent when to call which tool and how to navigate between workflows. This is the key to building effective agents — don't just give the agent tools, give it the right skills for the task.

File structure

skills/indy/
  SKILL.md                                    ← entry point, routing table
  skills.check-obligations.md                 ← "what needs doing?" — session start
  skills.categorize-transaction.md            ← core categorization loop
    └─ skills.categorize-transaction.edge-cases.md   ← personal payments, cash, non-EU, splits
  skills.receipts-and-manual-transactions.md  ← router for receipt/manual tx flows
    ├─ skills.receipts.upload-and-match.md           ← upload → OCR → match → link
    └─ skills.receipts.manual-transactions.md        ← create expense notes, cash tx
  skills.submit-vat-declaration.md            ← CA3/CA12 closing through teletransmission
  skills.end-of-year.md                       ← router for annual declarations
    ├─ skills.end-of-year.is-advance.md              ← IS advance payment (2571)
    ├─ skills.end-of-year.ag-pv.md                   ← shareholder meeting minutes
    └─ skills.end-of-year.other-declarations.md      ← DAS2, CFE, vehicle taxes
  skills.investigate-numbers.md               ← drill from statements to transactions
  skills.other.md                             ← router for read-only questions
    ├─ skills.other.financial-statements.md          ← bilan, P&L, trial balance
    ├─ skills.other.invoicing.md                     ← client invoices, KPIs, products
    └─ skills.other.company-info.md                  ← profile, company, associates, banks
  skills.context.md                           ← deep context for agent setup (NOT runtime)

Loading pattern

Skills are organized as parent files (routers, 20-70 lines) and sub-files (detail, 13-56 lines). An agent should:

1. Load SKILL.md → read the routing table
2. Load the parent for the current task → it tells you which sub-file to load
3. Load the sub-file when you hit a fork → do the work
4. Unload the sub-file when done → return to parent or check-obligations

Never load everything at once. Maximum context at any point should be ~170 lines (parent + one sub-file).

Cross-workflow jumps

Workflows are not linear. An agent frequently hits a fork and needs a different skill:

• Categorization → "paid from personal account" → skills.receipts.manual-transactions.md
• Categorization → "numbers look wrong" → skills.investigate-numbers.md
• VAT closing → uncategorized transactions → skills.categorize-transaction.md
• VAT closing → declaration amounts unexpected → skills.investigate-numbers.md
• End-of-year → AG/PV needs fiscal result → skills.other.financial-statements.md

The skill files document every jump point explicitly.

Creating Agents

This section is for an AI (or human) creating agents that use Indy tools.

The core idea: OS-level tool isolation

Every tool compiles to a standalone executable in bin/. Each is a self-contained ~27KB script that takes a JSON params string and an INDY_REFRESH_TOKEN env var. That's it — no runtime dependencies, no plugin framework needed.

# Any tool works standalone:
INDY_REFRESH_TOKEN="..." ./bin/indy-categorize-count.js
INDY_REFRESH_TOKEN="..." ./bin/indy-get-balance-sheet.js '{"year":"2025"}'
INDY_REFRESH_TOKEN="..." ./bin/indy-set-accounting-account.js \
  '{"transactionId":"abc","subdivisionId":"sub1","accountingAccountNumber":"625100"}'

When you create an agent, you copy only the binaries it needs into the agent's workspace. The agent can only execute what's physically present in its directory. No indy-submit-vat-teletransmission.js in the folder? The agent literally cannot submit a VAT declaration. This is enforced at the filesystem level — no allowlists to misconfigure, no config to override.

Setting up an agent workspace

# 1. Build all binaries from the plugin source
cd openclaw-indy
bun run build.ts              # → bin/ now has 65 .js files

# 2. Create the agent workspace
mkdir -p ~/agents/indy-categorize/tools
mkdir -p ~/agents/indy-categorize/skills

# 3. Copy ONLY the tools this agent needs
cp bin/indy-categorize-count.js      ~/agents/indy-categorize/tools/
cp bin/indy-list-pending-transactions.js ~/agents/indy-categorize/tools/
cp bin/indy-get-similar-transactions.js  ~/agents/indy-categorize/tools/
cp bin/indy-get-category-groups.js       ~/agents/indy-categorize/tools/
cp bin/indy-set-accounting-account.js    ~/agents/indy-categorize/tools/
cp bin/indy-set-transaction-vat.js       ~/agents/indy-categorize/tools/
# ... etc — only what's needed

# 4. Copy ONLY the skill files this agent needs
cp skills/indy/SKILL.md                                ~/agents/indy-categorize/skills/
cp skills/indy/skills.check-obligations.md             ~/agents/indy-categorize/skills/
cp skills/indy/skills.categorize-transaction.md        ~/agents/indy-categorize/skills/
cp skills/indy/skills.categorize-transaction.edge-cases.md ~/agents/indy-categorize/skills/
cp skills/indy/skills.receipts.upload-and-match.md     ~/agents/indy-categorize/skills/

# 5. The agent's workspace now looks like:
# ~/agents/indy-categorize/
#   tools/
#     indy-categorize-count.js
#     indy-list-pending-transactions.js
#     indy-get-similar-transactions.js
#     indy-get-category-groups.js
#     indy-set-accounting-account.js
#     indy-set-transaction-vat.js
#   skills/
#     SKILL.md
#     skills.check-obligations.md
#     skills.categorize-transaction.md
#     skills.categorize-transaction.edge-cases.md
#     skills.receipts.upload-and-match.md

The agent sees 6 tools and 5 skill files. It cannot access the other 59 tools. It cannot submit VAT declarations, delete transactions, or modify company settings — those binaries don't exist in its world.

Principles

1. One job per agent. A categorization agent shouldn't also do VAT teletransmission. Separate concerns into separate workspaces.
2. Skill files are the instructions. Tools without skills = an agent calling random endpoints. Always copy the matching skill files alongside the tools.
3. Read-only by default. Start with only read-tool binaries. Add write-tool binaries after the agent works correctly with reads. If a write tool isn't in the workspace, it can't be called.
4. skills.context.md is for setup, not runtime. Read it once to understand the domain when designing your agent. Don't copy it into the agent workspace — it's 250 lines of background context that wastes tokens at runtime.

Data conventions the agent system prompt must include

• Amounts: cents (÷100 for €). Negative = expense, positive = income.
• Dates: YYYY-MM-DD. Fiscal periods: YYYY-MM (e.g. 2026-01).
• VAT rates: basis points — 2000 = 20%, 1000 = 10%, 550 = 5.5%, 0 = exempt.
• subdivisions[].id is the subdivisionId needed for write operations.
• ⚠️ Always confirm with the user before any write operation.

Example 1: Daily Categorization Agent

The most common agent. Checks for uncategorized transactions, classifies them, attaches receipts.

Tools to copy (15 binaries):

# Read (11)
indy-list-pending-transactions.js
indy-categorize-count.js
indy-get-similar-transactions.js
indy-get-category-groups.js
indy-get-receipts.js
indy-get-receipt.js
indy-list-unpaired-receipts.js
indy-get-receipt-matching-transactions.js
indy-get-matching-receipts.js
indy-get-matching-invoices.js
indy-get-company-settings.js

# Write (4)
indy-set-accounting-account.js
indy-set-transaction-vat.js
indy-annotate-transaction.js
indy-upload-receipt.js

Skills to copy (5 files):

SKILL.md
skills.check-obligations.md
skills.categorize-transaction.md
skills.categorize-transaction.edge-cases.md
skills.receipts.upload-and-match.md

System prompt:

You are an accounting assistant for Indy.fr. Your job is to categorize
uncategorized bank transactions.

Start every session by calling indy-categorize-count. If 0, tell the user
everything is up to date. If > 0, call indy-list-pending-transactions and
work through them one at a time.

For each transaction:
1. Check indy-get-similar-transactions — reuse the same category if found
2. If no match, use indy-get-category-groups to find the right PCG code
3. Set the account with indy-set-accounting-account
4. Set VAT with indy-set-transaction-vat (most French expenses = 2000 = 20%)
5. Check for matching receipts and attach if found

Always confirm with the user before writing. When unsure about a category,
ask — do not guess. See your skill files for edge cases (personal payments,
cash, non-EU suppliers).

Setup:

AGENT=~/agents/indy-categorize
mkdir -p $AGENT/{tools,skills}
for f in indy-list-pending-transactions indy-categorize-count \
  indy-get-similar-transactions indy-get-category-groups \
  indy-get-receipts indy-get-receipt indy-list-unpaired-receipts \
  indy-get-receipt-matching-transactions indy-get-matching-receipts \
  indy-get-matching-invoices indy-get-company-settings \
  indy-set-accounting-account indy-set-transaction-vat \
  indy-annotate-transaction indy-upload-receipt; do
  cp bin/$f.js $AGENT/tools/
done
cp skills/indy/SKILL.md skills/indy/skills.check-obligations.md \
   skills/indy/skills.categorize-transaction.md \
   skills/indy/skills.categorize-transaction.edge-cases.md \
   skills/indy/skills.receipts.upload-and-match.md $AGENT/skills/

Example 2: Read-Only Financial Overview Agent

Cannot modify anything. Answers questions about finances, shows statements, checks obligations.

Tools to copy (30 binaries — all reads):

indy-get-user.js  indy-get-dashboard.js  indy-get-accounting-summary.js
indy-list-transactions.js  indy-get-transaction-filters.js  indy-get-transaction.js
indy-list-bank-accounts.js  indy-list-activities.js  indy-get-closing-state.js
indy-categorize-count.js  indy-list-unpaired-receipts.js  indy-get-vat-config.js
indy-get-balance-sheet.js  indy-get-income-statement.js  indy-get-general-balance.js
indy-get-period-choices.js  indy-get-company-settings.js  indy-get-associates.js
indy-list-client-invoices.js  indy-list-draft-invoices.js  indy-get-invoice-kpis.js
indy-list-products.js  indy-get-cfe-declaration.js  indy-get-das2.js
indy-get-is-advance-payment.js  indy-get-ag-pv.js  indy-get-vehicle-taxes.js
indy-list-documents.js  indy-list-invoices.js  indy-get-payment.js

Skills to copy (7 files):

SKILL.md
skills.check-obligations.md
skills.other.md
skills.other.financial-statements.md
skills.other.invoicing.md
skills.other.company-info.md
skills.investigate-numbers.md

System prompt:

You are a financial assistant for Indy.fr. You can read all accounting data
but cannot modify anything.

When the user asks a question:
- "What needs doing?" → indy-list-activities + indy-categorize-count
- "How's my year?" → indy-get-dashboard or indy-get-income-statement
- "Show me my balance sheet" → indy-get-period-choices then indy-get-balance-sheet
- "Any late invoices?" → indy-list-client-invoices with status LATE
- "Why is this number wrong?" → follow skills.investigate-numbers.md

Present amounts in euros (divide cents by 100). Use the skill files to know
which tools to call and in what order.

Zero write binaries in the workspace = zero risk of accidental modification.

Example 3: VAT Filing Agent

Has write access to closing steps and teletransmission. The most dangerous agent — only create this when you specifically need VAT filing.

Tools to copy (22 binaries):

# Read (17)
indy-list-activities.js  indy-get-vat-config.js  indy-get-closing-state.js
indy-refresh-closing.js  indy-get-categorize-all-status.js
indy-get-vat-rate-verification.js  indy-get-vat-company-identity.js
indy-get-vat-declaration-summary.js  indy-get-vat-credit-refund.js
indy-get-vat-validation.js  indy-get-pending-teletransmission.js
indy-get-last-teletransmission-state.js  indy-categorize-count.js
indy-list-unpaired-receipts.js  indy-get-company-settings.js
indy-get-general-balance.js  indy-list-transactions.js

# Write — closing steps (4)
indy-set-closing-step-status.js  indy-set-vat-company-identity.js
indy-validate-vat-declaration-summary.js  indy-validate-vat-credit-refund.js

# Write — IRREVERSIBLE (1)
indy-submit-vat-teletransmission.js

Skills to copy (4 files):

SKILL.md
skills.check-obligations.md
skills.submit-vat-declaration.md
skills.investigate-numbers.md

System prompt:

You are a VAT declaration assistant for Indy.fr. Walk the user through VAT
closing step by step.

CRITICAL: indy-submit-vat-teletransmission is IRREVERSIBLE. It sends the
declaration to French tax authorities. NEVER call it without explicit user
confirmation. Present the full declaration summary first.

Follow skills.submit-vat-declaration.md exactly. If uncategorized transactions
block the closing, tell the user to run the categorization agent first.

Security note: If you want an agent that can prepare a VAT declaration but not submit it, simply don't copy indy-submit-vat-teletransmission.js into the workspace. The agent can walk through all steps but physically cannot teletransmit.

Building your own agent

1. Read skills.context.md once to understand the domain (daily loop vs periodic loop, what matters, what's safe to automate). Do NOT copy this into the agent workspace.
2. Pick a job — categorize, file VAT, answer questions, process receipts, etc.
3. Select tools from the inventory above — only the binaries needed for that job.
4. Select skill files — the parent + relevant sub-files. Match skills to tools.
5. Create the workspace — mkdir -p agent/{tools,skills}, copy binaries + skills.
6. Write a system prompt — reference the skill files, explain the role, include data conventions.
7. Start read-only — omit write binaries until the agent works correctly with reads. Then add them one at a time.

Security model summary

Source (openclaw-indy/)
  bin/                      ← all 65 binaries (master copy)
  skills/indy/              ← all skill files (master copy)

Agent workspace (~/agents/my-agent/)
  tools/                    ← only the binaries this agent needs
  skills/                   ← only the skill files this agent needs

• The agent can only run tools present in its tools/ directory.
• No binary = no capability. Period.
• Write tools are opt-in by physical presence, not by configuration.
• To revoke a capability, rm the binary. To grant one, cp it in.
• Each agent is fully self-contained — you can tar the workspace and deploy it anywhere that has bun.

CLI Commands (OpenClaw plugin mode)

When installed as an OpenClaw plugin, these commands are available:

openclaw indy login [token]         # Setup authentication
openclaw indy status                 # Check login status
openclaw indy me                     # Show user info
openclaw indy transactions           # List transactions
openclaw indy bank                   # List bank accounts
openclaw indy receipts [txId]        # List receipts for a transaction
openclaw indy upload <txId> <file>   # Upload a receipt
openclaw indy dashboard [year]       # Show fiscal dashboard
openclaw indy fetch <path>           # Raw API request

Config Reference

plugins:
  entries:
    indy:
      enabled: true
      config:
        refreshToken: "..."                    # Required — Firebase refresh token
        firebaseApiKey: "..."                  # Optional — custom Firebase API key
        baseUrl: "https://app.indy.fr/api"     # Optional — API base URL

For standalone binaries, the equivalent config is environment variables:

export INDY_REFRESH_TOKEN="..."              # Required
export INDY_FIREBASE_API_KEY="..."           # Optional
export INDY_BASE_URL="https://app.indy.fr/api"  # Optional

License

MIT